A teaching resource group configuration and mistake question backtracking method based on knowledge point association

Abstract

The application provides a knowledge point association-based teaching resource grouping and mistake backtracking method, relates to the technical field of intelligent education and computer-aided teaching, and comprises the following steps: acquiring knowledge association data, teaching resource data and learning records of a target learner, generating a target problem-solving knowledge path according to problem-solving information of a target mistake, and matching a problem-solving process with the target problem-solving knowledge path to determine an abnormal problem-solving position; determining candidate error factors according to the abnormal problem-solving position and the learning records, respectively performing hypothetical state adjustment on the candidate error factors, and determining a target error factor set according to corresponding answer result prediction values; determining target teaching resources and a presentation sequence thereof based on the target error factor set, generating a teaching resource grouping sequence; providing a corresponding verification task after the sequence is completed, and updating an element mastery state and / or an association application state according to an answer result, so that error factor positioning, resource grouping, verification and learning state updating are realized.

Description

Technical Field

[0001] This application belongs to the field of smart education and computer-aided instruction technology, specifically involving a method for combining teaching resources and reviewing incorrect questions based on knowledge point association. Background Technology

[0002] With the development of online teaching, smart campuses, and digital question bank technologies, teaching platforms can continuously collect learners' answer data during exams, assignments, and self-study, and provide learners with explanation resources, practice questions, or review tasks based on their error records. By analyzing these errors and configuring corresponding teaching resources, it helps to identify learners' knowledge gaps and enable targeted learning interventions.

[0003] Existing teaching platforms typically tagged questions with labels such as chapter, knowledge point, question type, and difficulty. When a learner answers incorrectly, the system determines the knowledge point to be reviewed based on the corresponding knowledge point label, and then retrieves teaching resources with the same or similar labels from the question bank or resource library. Some systems also combine learning data such as historical accuracy rate, question difficulty, and answering time to filter or sort teaching resources.

[0004] However, for comprehensive questions that require the continuous use of multiple knowledge points or the connection of conditional judgments, expression transformations, and reasoning between different knowledge points, relying solely on the overall correct / incorrect result and knowledge point tags makes it difficult to determine at which specific step of the problem-solving process the error occurred. It also makes it difficult to distinguish whether the learner has not mastered a particular knowledge point or has failed to correctly connect and apply related knowledge under specific question conditions. Therefore, the resources selected may repeat content already mastered or fail to address the actual error-causing step. Furthermore, after learners complete the recommended resources, existing methods typically lack further verification of the original cause of the error, making it difficult to determine whether the error has been effectively corrected. Summary of the Invention

[0005] This application provides a method for combining teaching resources and reviewing incorrect questions based on knowledge point associations to solve one of the aforementioned technical problems.

[0006] The technical solution adopted in this application is as follows: This application provides a method for combining teaching resources and reviewing incorrect questions based on knowledge point associations, including: Acquire knowledge-related data, teaching resource data, and learning records of the target learner. The knowledge-related data includes multiple knowledge elements and the relationships between the knowledge elements. The learning records include the element mastery status corresponding to the knowledge elements and the related application status corresponding to the relationships. The element mastery status and the related application status constitute the learning status of the target learner. Based on the solution information of the target incorrect questions, a target solution knowledge path is generated, and the answer process of the target learner for the target incorrect questions is matched with the target solution knowledge path to determine the abnormal solution location; Based on the abnormal problem-solving location and the learning record, candidate error-causing factors are determined from the target problem-solving knowledge path. Hypothetical state adjustments are made to each candidate error-causing factor. Based on the learning state after the hypothetical state adjustment, the predicted value of the target learner's answer to the target wrong question is determined. Based on the predicted value of the answer corresponding to each candidate error-causing factor, the target error cause set is determined. Based on the target error set and the correspondence between teaching resources and knowledge elements and / or relationships in the teaching resource data, target teaching resources covering the target error set are determined from the teaching resource data. Based on the sequential dependency between knowledge elements and / or relationships corresponding to each target teaching resource, the presentation order of each target teaching resource is determined, and a teaching resource combination sequence is generated. The teaching resource combination sequence is provided to the target learner in sequence, and after the teaching resource combination sequence is completed, a verification task corresponding to the target error set is provided to the target learner; based on the verification object corresponding to the verification task and the target learner's answer to the verification task, the corresponding element mastery status and / or associated application status are updated.

[0007] According to one embodiment of this application, the knowledge-related data is represented by a directed association structure, the knowledge elements constitute knowledge nodes in the directed association structure, and the association relationships constitute directed association edges connecting different knowledge nodes; The relationships include at least one of the following: prerequisite relationship, condition triggering relationship, representation transformation relationship, reasoning inheritance relationship, and joint application relationship; For each knowledge node, a corresponding element mastery state is set, and for each directed edge, a corresponding association application state is set.

[0008] According to one embodiment of this application, generating the target problem-solving knowledge path based on the problem-solving information of the target incorrect question includes: Obtain the standard solution process corresponding to the target incorrect question, and divide the standard solution process into multiple solution steps arranged in the solution order; Determine the input state, problem-solving operation, and output state corresponding to each of the problem-solving steps, and determine the knowledge elements and / or relationships invoked when executing each of the problem-solving operations; According to the execution order of each problem-solving step, the problem-solving steps and the knowledge elements and / or relationships called by each problem-solving step are connected to generate the target problem-solving knowledge path.

[0009] According to one embodiment of this application, the step of matching the target learner's answer process for the target incorrect question with the target problem-solving knowledge path to determine the abnormal problem-solving location includes: Multiple learner problem-solving steps are extracted from the question-answering process. The learner problem-solving steps include intermediate answer results and answering operations used to obtain the intermediate answer results. Based on the consistency of input state, equivalence of answering operation, and consistency of output state between each learner's problem-solving step and each problem-solving step in the target problem-solving knowledge path, the learner's problem-solving steps are sequentially matched with each problem-solving step; The position of the first learner's problem-solving step that does not meet the preset matching conditions in the target problem-solving knowledge path is determined as the abnormal problem-solving position.

[0010] According to one embodiment of this application, determining candidate error-causing factors from the target problem-solving knowledge path based on the abnormal problem-solving location and the learning record includes: Starting from the abnormal solution location, backtrack along the reverse execution direction of the target solution knowledge path to determine the solution steps corresponding to the abnormal solution location, as well as the knowledge elements and relationships upstream of the solution steps; The knowledge elements and relationships obtained through backtracking are identified as the initial error-causing factors; Extract historical correct answers and historical incorrect answers corresponding to each of the initial error-causing factors from the learning records. Filter each of the initial error-causing factors based on the historical correct answers and historical incorrect answers to obtain the candidate error-causing factors.

[0011] According to one embodiment of this application, the step of performing hypothetical state adjustments on each of the candidate error-causing factors and determining the target error cause set based on the predicted answer values ​​corresponding to each of the candidate error-causing factors includes: Based on the learning state before hypothetical state adjustment, determine the baseline predicted value of the target learner's answer to the target incorrect question; When the candidate error-causing factor is a knowledge element, the mastery status of the element corresponding to the candidate error-causing factor is temporarily adjusted to the preset mastery status, while keeping the mastery status and associated application status of other elements unchanged. When the candidate error-causing factors are related, the associated application state corresponding to the candidate error-causing factors is temporarily adjusted to the preset application state, while keeping other associated application states and element mastery states unchanged. Based on the learning state after each hypothetical state adjustment, the corresponding predicted value of the answer result is determined, and the target error cause set is determined from the candidate error-causing factors according to the change of each predicted value of the answer result relative to the baseline predicted value of the answer result.

[0012] According to one embodiment of this application, the teaching resource data also includes knowledge element coverage information, correlation relationship coverage information, resource function type, resource difficulty and expected learning time for each teaching resource; The step of determining the target teaching resources covering the target error set based on the target error set includes: Based on the knowledge element coverage information and the relationship coverage information, candidate teaching resources that can cover each target error in the target error set are determined; The target teaching resource is determined from the candidate teaching resources based on the difficulty of the resource, the expected learning time, and the degree of content overlap between different candidate teaching resources. Wherein, when the target error set includes correlations, the target teaching resources include at least one correlation bridging resource for explaining the correlations.

[0013] According to one embodiment of this application, the generation of the teaching resource combination sequence includes: Based on the prerequisite relationships between the knowledge elements and related relationships covered by each of the target teaching resources, the knowledge dependency relationships between the target teaching resources are determined. Based on the resource function type of each of the target teaching resources, determine the functional connection relationship between each of the target teaching resources; Based on the knowledge dependency relationship and the functional connection relationship, the target teaching resources are sorted to generate a teaching resource combination sequence that includes at least two of the following: knowledge explanation resources, association bridging resources, comparison and analysis resources, and transfer practice resources.

[0014] According to one embodiment of this application, the verification task corresponding to the target error set includes at least one of knowledge element verification items and related application verification items; The knowledge element verification item is used to verify the target learner's mastery of the knowledge elements in the target error set; The knowledge elements and relationships invoked by the verification item are the same as those invoked by the target incorrect question, and the verification item differs from the target incorrect question in at least one of the following: question object, question value, language expression, and application scenario.

[0015] According to one embodiment of this application, updating the corresponding element mastery status and / or associated application status based on the verification object corresponding to the verification task and the target learner's response to the verification task includes: When the answer to the knowledge element verification item meets the preset pass condition, and the answer to the related application verification item does not meet the preset pass condition, the element mastery status corresponding to the knowledge element verification item is maintained, and the related application status corresponding to the related application verification item is reduced. When the answer results of the knowledge element verification item and the related application verification item do not meet the preset pass conditions, the corresponding element mastery status and related application status are reduced respectively. When the answer to the related application verification item meets the preset pass condition, the related application status corresponding to the related application verification item is improved, and the teaching effectiveness of the teaching resource combination sequence on the related relationship is recorded.

[0016] Due to the adoption of the above technical solution, the beneficial effects achieved by this application are as follows: This application maintains the mastery status of each knowledge element and the application status of the relationships between knowledge elements separately, and matches the learner's answer process with the target problem-solving knowledge path corresponding to the incorrect answers, thus locating abnormal problem-solving positions in the answer process. Based on this, it identifies candidate error-causing factors by combining learning records, performs hypothetical status adjustments on each candidate error-causing factor, and determines the target error set based on the adjusted predicted answer results. This allows for the differentiation between different error scenarios—lack of mastery of the knowledge element itself and failure to correctly apply the relationships between knowledge elements—reducing the bias caused by making general error judgments based solely on the overall correctness of the question. Furthermore, it selects data from teaching resource data to cover the target error causes. The set of target teaching resources is organized, and the presentation order of each target teaching resource is determined according to the sequential dependencies between corresponding knowledge elements and relationships. This allows resources such as knowledge explanation, connection bridging, comparison and analysis, and transfer exercises to be applied to the target error causes in sequence according to the knowledge processing logic, reducing the allocation of irrelevant and redundant resources. After the teaching resource combination sequence is completed, verification tasks corresponding to the target error cause set are provided, and the element mastery status and / or related application status are updated according to the answers of different verification objects. This forms a closed loop of error cause identification, resource combination, learning execution, targeted verification, and status update, making the learning status used in subsequent error analysis and teaching resource combination more consistent with the actual mastery of the target learners. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 System architecture diagram provided for embodiments of this application; Figure 2 A flowchart illustrating the teaching resource allocation and error correction method based on knowledge point association provided in this application embodiment. Detailed Implementation

[0018] The technical solution of this application will be further described below with reference to the accompanying drawings and embodiments. It should be understood that the following embodiments are only used to illustrate this application and are not intended to limit the scope of protection of this application.

[0019] In the description of the embodiments of this application, the terms "first," "second," etc., are used only to distinguish different objects and should not be construed as indicating or implying relative importance, nor are they used to limit the number of objects or the order of execution; the term "multiple" refers to two or more; the term "and / or" is used to describe the association relationship between related objects, indicating that any one or any combination of the listed related objects may exist. Unless otherwise expressly defined, the terms "acquire," "input," "output," "connect," "transmit," "call," etc., should be interpreted broadly.

[0020] This application's embodiments can be applied to smart campus platforms, online learning platforms, online examination platforms, school-based resource platforms, or other computer systems capable of collecting learners' answer data and providing learning resources. The method can be executed by a teaching server, or it can be executed collaboratively by the teaching server, student terminals, teacher terminals, and a data server.

[0021] Existing smart campus systems typically include functions such as question bank management, resource management, knowledge point management, and error correction improvement. Knowledge points can be organized into root knowledge points and sub-knowledge points, and the error correction improvement function can push relevant questions based on the knowledge points involved in the incorrect questions. However, simply selecting questions based on the knowledge points corresponding to the incorrect questions makes it difficult to distinguish whether learners have not mastered individual knowledge content or have mastered the relevant knowledge content but cannot correctly apply the relationships between the knowledge content during the problem-solving process.

[0022] Existing teaching platforms can also retrieve and combine questions and resources according to version, volume, chapter, and knowledge point. The multi-level category configuration method used by e-commerce platforms can also be used to construct a multi-level classification index for teaching resources. This application does not simply combine resources according to the above-mentioned classification tags, but after identifying the knowledge elements or relationships that lead to the target wrong questions, it ensures that the selected teaching resources cover the identified target error causes, and organizes the presentation order of resources according to knowledge dependencies.

[0023] To ensure the traceability of the diagnostic and status update processes, the system can store configuration parameters, business data, and runtime logs separately. Configuration parameters are used to store knowledge association types, status judgment conditions, and resource filtering conditions, while runtime logs are used to store the data used in the current diagnosis, output results, and status update records. In a similar data architecture, parameter information and runtime logs can be stored separately to facilitate the tracking of module runtime status and processing.

[0024] Knowledge elements refer to units of knowledge or problem-solving abilities that can be independently accessed or evaluated when answering teaching questions. In different teaching scenarios, knowledge elements can be concepts, properties, theorems, rules, formulas, calculation methods, problem-solving skills, or other content with independent teaching significance. For example, in the scenario of solving product-form equations, knowledge elements may include the property of zero product and methods for solving linear equations in one variable.

[0025] Relationships refer to the directional or dependent relationships formed between different knowledge elements during the learning or problem-solving process. Relationships can include at least one of the following: prerequisite relationships, conditional triggering relationships, representational transformation relationships, reasoning succession relationships, and joint application relationships.

[0026] The prerequisite relationship indicates that one knowledge element is a prerequisite for understanding or using another knowledge element; the conditional trigger relationship indicates that the corresponding knowledge element needs to be invoked when specific problem conditions are met; the representation transformation relationship indicates that a transformation needs to be completed between different forms of expression; the reasoning succession relationship indicates that the processing result of the previous knowledge element serves as the input of the next knowledge element; the joint application relationship indicates that multiple knowledge elements need to be used in combination in the same problem-solving step or in consecutive problem-solving steps.

[0027] The mastery status of knowledge elements refers to the degree to which the learner has mastered the knowledge elements themselves; the application status refers to the degree to which the learner can identify, recall, and apply the relationships between knowledge elements when the corresponding application conditions are met. The mastery status of knowledge elements and the application status of knowledge elements are stored and maintained separately to avoid judging all related knowledge elements as unmastered due to an incorrect answer to a comprehensive question.

[0028] The state of element mastery and the state of related application can be represented by discrete or continuous states. For example, discrete states can include unmastered, pending verification, and mastered; continuous states can be represented by state values ​​within a preset range. The state representation method and the transition conditions between different states can be configured before the system runs. The current error diagnosis uses the state stored at the start of the diagnosis; hypothetical state adjustments made during the diagnosis process are not directly written back to the original state.

[0029] A target problem-solving knowledge path refers to an ordered data structure formed based on the problem-solving information of target incorrect questions. This structure represents the problem-solving steps and the knowledge elements and relationships invoked by each step. A target problem-solving knowledge path can be a linear path or include multiple optional branches. When a problem has multiple equivalent solutions, multiple candidate problem-solving knowledge paths can be configured.

[0030] An abnormal problem-solving location refers to the point where the learner's answer process does not meet the preset matching conditions with the target problem-solving knowledge path. In a preferred implementation, the abnormal problem-solving location is the position corresponding to the first problem-solving step that does not meet the preset matching conditions.

[0031] Candidate error-causing factors refer to knowledge elements or relationships that are related to the abnormal solution location and may lead to an error in the current answer; the "target error-causing factor set" refers to one or more target error-causing factors identified from the candidate error-causing factors after hypothetical state adjustment and answer result prediction.

[0032] Hypothetical state adjustment refers to temporarily adjusting the mastery or application status of elements corresponding to candidate error-causing factors without modifying the learner's actual stored state, and then using the temporarily adjusted state to re-predict the answer to the target incorrect question. Each hypothetical state adjustment can be performed in a temporary copy of the learning state, and the temporary copy is deleted or restored after evaluation.

[0033] The predicted answer result refers to the result of predicting whether a learner can correctly complete the target incorrect questions based on their mastery and application of the elements invoked in the target problem-solving knowledge path. The predicted value can be the probability of correct answer, the predicted score, the correct answer level, or other data that reflects the predicted answer result.

[0034] A teaching resource combination sequence refers to a sequence of resources formed by selecting multiple teaching resources based on the target set of error factors and determining the presentation order of these resources. This resource sequence is not an unordered collection of resources, but rather an executable sequence of resources organized according to knowledge dependencies and / or resource functional connections.

[0035] Verification tasks are questions or learning tasks used to verify whether the target learners have eliminated the root causes of errors after completing the sequence of instructional resources. Verification tasks can include knowledge element verification items and related application verification items. Knowledge element verification items primarily verify the knowledge elements themselves, while related application verification items primarily verify whether the relationships between knowledge elements can be correctly invoked.

[0036] Figure 1 The system architecture provided in an embodiment of this application is illustrated. For example... Figure 1 As shown, the system may include student terminals, teacher terminals, a teaching server, and a data storage device. Student terminals and teacher terminals communicate with the teaching server via wired or wireless networks, respectively, and the teaching server is connected to the data storage device.

[0037] The student terminal is used to display teaching questions, collect the learners' answer processes, output teaching resource combination sequences and verification tasks, and transmit the learners' operation results to the teaching server. The student terminal can be a personal computer, tablet computer, smartphone, or a teaching terminal with interactive display functions.

[0038] The teacher terminal is used to configure knowledge-related data, review standard problem-solving processes, configure knowledge elements and relationships corresponding to teaching resources, and view low-confidence diagnostic results. The teacher terminal is not a separate piece of hardware required to execute the methods described in this application; the teacher configuration function can also be integrated into the management console of the teaching server.

[0039] The teaching server may include modules for knowledge association data management, learning record management, problem-solving path processing, answer process matching, error analysis, teaching resource allocation, verification task processing, and status update. These modules can function as software units running on the same server or be deployed on multiple servers.

[0040] The knowledge association data management module is used to establish, read, and maintain knowledge elements and the relationships between them. The learning record management module is used to read and store the learner's historical answer records, current answering process, element mastery status, and application status.

[0041] The problem-solving path processing module generates a target problem-solving knowledge path based on the standard problem-solving process of the target incorrect questions. The answer process matching module matches the target learner's answer process with the target problem-solving knowledge path and outputs the location of abnormal solutions.

[0042] The error analysis module is used to identify candidate error-causing factors based on the abnormal solution location, and to perform hypothetical state adjustments on the candidate error-causing factors in a temporary copy of the learning state to obtain the predicted answer value corresponding to each candidate error-causing factor.

[0043] The teaching resource allocation module is used to filter target teaching resources from the teaching resource data based on the target error set and determine the presentation order of the target teaching resources. The verification task processing module is used to provide verification tasks to target learners after resource learning is completed, and the status update module is used to update the element mastery status and / or related application status based on the answer results of the verification task.

[0044] The data storage device can include a knowledge association database, a question database, a teaching resource database, a learning record database, and an operation log database. These databases can be independent data tables or logical databases, or they can be stored in the same database system.

[0045] A knowledge association database can record knowledge element identifiers, knowledge element names, subjects, grade levels, association relationship identifiers, starting knowledge elements, target knowledge elements, relationship types, and relationship directions. A teaching resource database can record teaching resource identifiers, resource content addresses, knowledge element coverage information, association relationship coverage information, resource function types, resource difficulty, estimated learning time, and resource versions.

[0046] The learning record database can record learner identifiers, question identifiers, answer time, answer results, answer process, knowledge elements and relationships invoked by the question, as well as the mastery and application status of elements before and after answering. The operation log database can record the current diagnostic version, candidate error-causing factors, results of each hypothetical adjustment, target error set, teaching resource combination sequence, and verification results.

[0047] The data interaction sequence between modules is as follows: the student terminal sends the answer process of the target incorrect question to the learning record management module; the problem-solving path processing module reads the solution information of the target incorrect question from the question database and the knowledge association database and generates the target problem-solving knowledge path; the answer process matching module determines the abnormal solution location based on the answer process and the target problem-solving knowledge path; the error analysis module reads the learning record and determines the target error cause set; the teaching resource combination module reads the teaching resource data and generates the teaching resource combination sequence; the student terminal executes the teaching resource combination sequence and the verification task; the status update module updates the status based on the verification result. The updated status is used for subsequent error diagnosis or subsequent practice and is not used as the pre-input for this hypothetical status adjustment.

[0048] Figure 2 This illustrates a method for combining teaching resources and reviewing incorrect questions based on knowledge point association, provided in an embodiment of this application. For example... Figure 2 As shown, the method may include steps S100 to S500.

[0049] S100, acquire knowledge association data, teaching resource data, and learning records of the target learner. The knowledge association data includes multiple knowledge elements and the relationships between the knowledge elements. The learning records include the element mastery status corresponding to the knowledge elements and the association application status corresponding to the association relationships. The element mastery status and the association application status constitute the learning status of the target learner.

[0050] Knowledge-related data can be retrieved from the knowledge-related database by the knowledge-related data management module. This data includes multiple knowledge elements and the relationships between them. Teaching resource data can be retrieved from the teaching resource database by the teaching resource assembly module. The learning records of the target learners can be retrieved from the learning record database by the learning record management module.

[0051] In one possible implementation, knowledge-related data is represented by a directed association structure. Each knowledge element constitutes a knowledge node in the directed association structure, and the relationships between knowledge elements constitute directed association edges connecting different knowledge nodes. The direction of the directed association edges indicates the direction of invocation, transformation, or dependency between knowledge elements.

[0052] For example, knowledge element A is "zero product property", knowledge element B is "solving linear equations in one variable", and the association relationship E is "converting a product-form equation into multiple linear equations in one variable based on the zero product property". In a directed association structure, knowledge elements A and B can be configured as knowledge nodes, and the association relationship E can be configured as a directed association edge associated with the two knowledge nodes.

[0053] For each knowledge node, the system maintains the learner's mastery status of the corresponding element; for each directed edge, the system maintains the learner's application status of the corresponding edge. For learners with existing historical learning records, the system can generate or update the corresponding status based on historical correct and incorrect answers. For learners lacking historical learning records, the system can use a pre-configured initial status and update it after obtaining subsequent answer data.

[0054] Each teaching resource in the teaching resource data can be indexed by subject, grade, textbook version, volume, chapter, knowledge element, relationship, and resource type. Each teaching resource has knowledge element coverage information and / or relationship coverage information, which are used to indicate the knowledge elements and relationships that the teaching resource can explain, demonstrate, train, or verify.

[0055] Teaching resources can include text explanations, image explanations, audio and video resources, interactive demonstrations, example questions, analytical questions, practice questions, and verification questions. The knowledge element coverage information and relational coverage information can be configured by teachers or resource maintainers when resources are added to the database, or confirmed by resource reviewers based on the resource content.

[0056] The learning record of the target learner should include at least the learner's current mastery and application of the elements. In one possible implementation, the learning record should also include historical question identifiers, answer time, answer result, answer process, question difficulty, and the knowledge elements and relationships invoked by the question, so as to subsequently filter records related to candidate error-causing factors from the historical record.

[0057] Step S100 outputs the knowledge association data, teaching resource data, and learning status snapshot used in this diagnosis. The learning status snapshot remains unchanged during the current error analysis, while hypothetical state adjustments are performed in its temporary copy. This avoids premature modification of the actual learning status before the current diagnosis is completed.

[0058] S200: Generate a target problem-solving knowledge path based on the problem-solving information of the target incorrect question, and match the target learner's answer process for the target incorrect question with the target problem-solving knowledge path to determine the abnormal problem-solving location.

[0059] Targeted incorrect questions can be those questions that the target learner answered incorrectly in exams, assignments, self-study, or historical exercises. The solution information for these targeted incorrect questions can include the question content, the standard answer, the standard solution process, the sequence of solution steps, and the knowledge elements and their relationships involved in each solution step.

[0060] The problem-solving path processing module obtains the standard solution process corresponding to the target incorrect problem and divides the standard solution process into multiple solution steps arranged in the solution order. Each solution step can be configured with input status, solution operation, and output status.

[0061] The input state represents the problem state or intermediate result obtained before executing the current problem-solving step; the problem-solving operation represents the recognition, transformation, calculation, or reasoning operation performed in the current step; the output state represents the intermediate result or final result obtained after executing the problem-solving operation.

[0062] The problem-solving path processing module further determines the knowledge elements and / or relationships called by each problem-solving step, and connects them according to the execution order of the problem-solving steps to generate the target problem-solving knowledge path.

[0063] For example, for the target incorrect question "Solve" The target problem-solving knowledge path can include: identifying equations in product form; applying the property of zero product; converting a product equal to zero into two factors equal to zero; solving the two linear equations in one variable; and outputting the solution. Among these, the "property of zero product" is a knowledge element, and "converting from product form to multiple linear equations in one variable" is an association.

[0064] When a target incorrect question has multiple equivalent solutions, multiple candidate solution knowledge paths can be pre-configured. The answer process matching module can match the target learner's answer process with multiple candidate solution knowledge paths, and select the candidate solution knowledge path with the highest overall matching degree as the target solution knowledge path.

[0065] The learner's answering process can be captured by the student terminal during the answering process. The answering process may include multiple learner problem-solving steps, each learner's problem-solving step includes intermediate answer results and answering operations used to obtain intermediate answer results.

[0066] Intermediate responses can be intermediate expressions, calculation results, selection results, or step-by-step answers entered by the learner; responses can include input, deletion, modification, selection, dragging, submission, or invoking prompts. Different operations can be accompanied by task identifiers and time information for the same response task to maintain the sequence of operations.

[0067] The answer process matching module performs sequential matching based on the consistency of input state, equivalence of answer operations, and consistency of output state between the learner's problem-solving steps and the problem-solving steps in the target problem-solving knowledge path.

[0068] Input state consistency is used to determine whether the intermediate result before the learner performs the current operation matches the input conditions of the standard steps; response operation equivalence is used to determine whether the operation used by the learner matches the standard operation or the pre-configured equivalent operation; output state consistency is used to determine whether the intermediate result obtained by the learner after the operation matches the output result of the standard steps.

[0069] When the input state, response operation, and output state of a learner's problem-solving step all meet the corresponding preset matching conditions, the matching of the next problem-solving step continues; when at least one of them does not meet the preset matching conditions, the position of the learner's problem-solving step in the target problem-solving knowledge path is determined as an abnormal problem-solving position.

[0070] In a preferred implementation, the first position that does not meet the preset matching conditions is identified as the abnormal solution position. Using the first abnormal position can reduce the interference of subsequent chain errors caused by preceding errors on the error cause judgment.

[0071] Taking the product-form equation as an example again, if the learner can recognize the problem as an equation problem but fails to convert the product being equal to zero into two factors being equal to zero, the abnormal solution point is the step of calling the zero product property and completing the form conversion, rather than the subsequent steps of solving a linear equation in one variable.

[0072] When the answer process data is incomplete, the steps that can be determined can be matched based on the collected intermediate results and operation records. If the existing data is insufficient to determine the location of the abnormal solution, the current diagnosis can be marked as pending verification, and a step-by-step diagnostic task can be provided first; after obtaining the supplementary answer process, the subsequent error cause analysis can be performed to avoid directly modifying the learning status based on a single final answer.

[0073] Step S200 outputs the location of the abnormal solution and the corresponding solution steps, which are then used by step S300 to determine the candidate error-causing factors.

[0074] S300, based on the abnormal problem-solving location and the learning record, determine candidate error-causing factors from the target problem-solving knowledge path; perform hypothetical state adjustments on each candidate error-causing factor, and determine the predicted value of the target learner's answer to the target wrong question based on the learning state after the hypothetical state adjustment; determine the target error cause set based on the predicted value of the answer corresponding to each candidate error-causing factor.

[0075] The error analysis module receives the location of the abnormal solution and backtracks along the reverse execution direction of the target solution knowledge path from that location. The backtracking objects include the solution steps corresponding to the abnormal solution location, as well as the knowledge elements and relationships upstream of that solution step.

[0076] The knowledge elements and relationships located upstream refer to those whose processing results, invocation results, or transformation results can serve as input conditions for abnormal problem-solving steps. The error cause analysis module identifies the knowledge elements and relationships obtained through backtracking as the initial error-causing factors.

[0077] For example, when the abnormal solution location is the step of "converting the product equal to zero into two linear equations in one variable", the initial error-causing factors may include the property of zero product, the ability to identify the product form, the correlation between the property of zero product triggering the equation splitting, and the solution method of linear equations in one variable that needs to be used subsequently.

[0078] The error analysis module extracts historical correct and incorrect answers from the learner's learning records, corresponding to each initial error-causing factor. The correspondence between historical records and initial error-causing factors can be determined based on the knowledge elements and relationships marked on the historical questions.

[0079] Historical correct answers demonstrate that the learner has previously used the relevant knowledge elements or relationships correctly, while historical incorrect answers demonstrate that the learner has previously made mistakes regarding the relevant knowledge elements or relationships. The system can further filter historical records based on answer time, question difficulty, question structure similarity, and the completeness of the answer process.

[0080] When a certain knowledge element has multiple recent correct answers, while a comprehensive question involving the relationship between that knowledge element and another knowledge element has multiple incorrect answers, the relationship can be retained as a candidate error-causing factor, and the priority of considering the knowledge element itself as a candidate error-causing factor can be reduced.

[0081] When an initial error-causing factor has neither available historical records nor sufficient evidence to be obtained from the current answering process, it can be retained as a factor to be verified, or evidence can be supplemented through subsequent diagnostic tasks. After filtering by historical records, candidate error-causing factors are obtained.

[0082] After identifying candidate error-causing factors, the error cause analysis module performs hypothetical state adjustments for each candidate factor. Each hypothetical state adjustment is performed based on an independent copy of the same state snapshot, without using the results of the previous hypothetical state adjustment as input for the next adjustment.

[0083] When the candidate error-causing factor is a knowledge element, the mastery status of the element corresponding to that knowledge element is temporarily adjusted to the preset mastery status, while the mastery status and related application status of other elements remain unchanged.

[0084] When a candidate error-causing factor is a correlation, the associated application status corresponding to that correlation is temporarily adjusted to the preset application status, while keeping other associated application statuses and element control statuses unchanged.

[0085] The preset mastery state and preset application state can be configured before the system runs, indicating that the corresponding knowledge elements or relationships can be used correctly. The state can be determined based on historical learning data, teaching requirements, or system status levels, and the current diagnosis uses the already configured state conditions.

[0086] Before making hypothetical state adjustments for candidate error-causing factors, the error cause analysis module determines the baseline predicted value of the target learner's answer to the target wrong question based on the original learning state.

[0087] In one possible implementation, the answer prediction module pre-configures the necessary knowledge state conditions for each problem-solving step in the target problem-solving knowledge path, and searches for pre-established result mapping data based on whether each necessary state meets the conditions, the number of risk steps, and the position of the first risk step, and outputs the predicted answer result value. The result mapping data is established before the current diagnosis begins.

[0088] After performing hypothetical state adjustments for each candidate error-causing factor, the same processing method as for determining the baseline answer prediction value is adopted to determine the corresponding answer prediction value based on the learning state after the hypothetical state adjustment.

[0089] The error cause analysis module compares the predicted response values ​​corresponding to each candidate error-causing factor. In a preferred implementation, it also compares the change of each predicted response value relative to the baseline predicted response value.

[0090] When the hypothetical state adjustment of a candidate error-causing factor causes the predicted value of the answer to meet a preset improvement condition, the candidate error-causing factor is added to the target error cause set. The preset improvement condition can be that the adjusted predicted value reaches a preset correct condition, or that the change in the predicted value before and after the adjustment reaches a preset change condition.

[0091] When multiple candidate error-causing factors meet the preset improvement conditions, all candidate error-causing factors can be added to the target error-causing factor set. When only one candidate error-causing factor meets the preset improvement conditions, the target error-causing factor set only includes that candidate error-causing factor.

[0092] When none of the candidate error-causing factors meet the preset improvement conditions, the candidate error-causing factor with the largest predicted value improvement can be marked as an error-causing factor to be verified, and diagnostic evidence can be supplemented through verification tasks; before obtaining supplementary evidence, the actual state of other knowledge elements and relationships should not be reduced.

[0093] For example, for the above product-form equations, candidate error-causing factors include the zero-product property, the method of solving linear equations in one variable, and the correlation between the product equal to zero and the transformation into multiple linear equations in one variable. If temporarily improving the mastery status of the first two knowledge elements results in a small change in the predicted value, while temporarily improving the application status of the correlation results in a significant increase in the predicted value, then the correlation is identified as the target error cause.

[0094] The hypothetical state adjustment in step S300 is only used for error cause analysis and does not modify the actual state in the learning record database. Step S300 outputs the target error cause set for step S400 to select target teaching resources.

[0095] S400: Based on the target error set and the correspondence between teaching resources and knowledge elements and / or relationships in the teaching resource data, determine the target teaching resources covering the target error set from the teaching resource data; based on the sequential dependency between the knowledge elements and / or relationships corresponding to each target teaching resource, determine the presentation order of each target teaching resource, and generate a teaching resource combination sequence.

[0096] The teaching resource allocation module determines the target teaching resources that cover the target set of error causes from the teaching resource data, based on the target set of error causes and the correspondence between teaching resources and knowledge elements and / or relationships in the teaching resource data.

[0097] In one possible implementation, the teaching resource data also includes information on knowledge element coverage, information on relational coverage, resource function type, resource difficulty, expected learning duration, and content characteristics.

[0098] Knowledge element coverage information is used to represent the knowledge elements that the teaching resources can explain or train; relationship coverage information is used to represent the relationships that the teaching resources can explain, demonstrate or train; resource function type is used to represent the function of the teaching resources in the learning process.

[0099] Resource functions can include knowledge explanation resources, association and bridging resources, comparative analysis resources, and transfer practice resources. Knowledge explanation resources are used to explain the knowledge elements themselves; association and bridging resources are used to explain why different knowledge elements are related, under what conditions they are related, and how to complete the corresponding transformation; comparative analysis resources are used to distinguish between situations where the association relationship is applicable and inapplicable; transfer practice resources are used to train the application of the same knowledge structure when the surface conditions change.

[0100] The teaching resource allocation module first determines candidate teaching resources that can cover each target error in the target error set based on the knowledge element coverage information and the correlation coverage information.

[0101] When the target error set includes knowledge elements, the candidate teaching resources should include resources that cover the corresponding knowledge elements; when the target error set includes relationships, the target teaching resources should include at least one relationship bridging resource that covers the relationships.

[0102] When multiple candidate teaching resources cover the same error cause, they can be selected based on the difficulty of the resources, the expected learning time, and the degree of content repetition between different candidate teaching resources.

[0103] The difficulty level of the resources can be determined based on the pre-configured difficulty level and adapted to the difficulty of the target incorrect questions, the learner's current status, and the teaching stage. The estimated learning time can be configured by the resource maintainer when the resources are added to the database, or it can be obtained based on statistics from historical learning records.

[0104] Content duplication can be compared based on knowledge element tags, relationship tags, resource function types, and content feature information of different resources. When two resources cover the same knowledge elements, the same relationships, and have the same resource function, the resource with a more suitable difficulty level or a shorter expected learning time can be prioritized to reduce ineffective duplication.

[0105] After identifying the target teaching resources, the presentation order of each target teaching resource is determined based on the sequential dependencies between the knowledge elements and / or relationships corresponding to each target teaching resource.

[0106] For example, when the first target teaching resource covers the knowledge elements of prerequisite knowledge, and the second target teaching resource covers the relationship between the prerequisite knowledge elements and the target knowledge elements, the first target teaching resource is placed before the second target teaching resource.

[0107] In one possible implementation, the functional connection relationship is also determined based on the resource function type of each target teaching resource. For example, resources can be organized in the order of knowledge explanation, connection and bridging, comparison and analysis, and transfer exercises.

[0108] The order of resource functions is not fixed. When the target learner has already reached the preset mastery level of relevant knowledge elements, and the target error is only related, knowledge explanation resources can be omitted, and the learning can start with the associative bridging resources, followed by comparative analysis resources and transfer practice resources.

[0109] For example, when the target error is "the relationship between the conversion from a product equal to zero to two linear equations in one variable", you can choose in sequence: related bridging resources that explain the triggering process of the zero product condition; comparative analysis resources to distinguish between "the product equals zero" and "the sum equals zero"; and transfer practice resources that convert different product expressions into linear equations in one variable.

[0110] The resulting sequence of teaching resources includes at least two of the following: knowledge explanation resources, linking and bridging resources, comparative analysis resources, and transfer practice resources. When the set of target error causes includes multiple target error causes with a sequential relationship, the resources corresponding to the upstream target error causes are completed first, and then the resources corresponding to the downstream target error causes are executed.

[0111] Step S400 outputs a teaching resource combination sequence, which can record resource identifiers, resource order, covered objects, estimated learning time and completion conditions, for students to output sequentially on their terminals.

[0112] S500, the teaching resource combination sequence is provided to the target learner in sequence, and after the teaching resource combination sequence is completed, a verification task corresponding to the target error set is provided to the target learner; according to the verification object corresponding to the verification task and the target learner's answer to the verification task, the corresponding element mastery status and / or associated application status are updated.

[0113] The teaching server sends the teaching resource combination sequence to the student terminals of the target learners. The student terminals display or run each target teaching resource in sequence according to the resource order, and record the execution status of each target teaching resource.

[0114] The execution status of a resource can include not started, in progress, and completed. The completion conditions for different resources can be determined based on the resource type. For example, a video resource can be marked as completed after a preset viewing condition is met; an interactive resource can be marked as completed after a specified operation is performed; and a practice resource can be marked as completed after submission.

[0115] Once the completion conditions for the current target teaching resource are met, the student terminal or teaching server opens the next target teaching resource. This ensures that teaching resources are executed in the order determined in step S400.

[0116] After the teaching resource assembly sequence is completed, the verification task processing module provides the target learner with verification tasks corresponding to the target set of error causes. Verification tasks can be selected from the question database or generated based on pre-configured question templates.

[0117] The verification task includes at least one of the following: knowledge element verification items and related application verification items. The knowledge element verification items are used to verify whether the target learner has mastered the knowledge elements in the target error set; the related application verification items are used to verify whether the target learner can correctly apply the relationships in the target error set during the specific problem-solving process.

[0118] The knowledge elements and relationships invoked by the verification item are the same as those invoked by the target incorrect question, and differ from the target incorrect question in at least one of the following: question object, question value, language expression, and application scenario.

[0119] For example, the target incorrect question is "solve". The verification item can be another product-form equation. The verification item has the same knowledge elements and relationships as the target incorrect question, but uses different values ​​or different expressions to reduce the possibility that the target learner will only memorize the answers based on the surface content of the target incorrect question.

[0120] The verification task processing module records the verification objects corresponding to each verification item. The verification object corresponding to the knowledge element verification item is one or more knowledge elements; the verification object corresponding to the association application verification item is one or more association relationships.

[0121] The status update module updates the corresponding element mastery status and / or related application status based on the verification object and the target learner's answer results for the verification task.

[0122] In one possible implementation, when the answer to the knowledge element verification item meets the preset pass condition, but the answer to the related application verification item does not meet the preset pass condition, it indicates that the target learner can independently understand the corresponding knowledge element, but still cannot correctly use the relationship between knowledge elements. Therefore, the mastery status of the corresponding knowledge element is maintained, while the related application status of the corresponding relationship is reduced or maintained.

[0123] When the answers to both the knowledge element verification item and the related application verification item fail to meet the preset pass conditions, it indicates that the target learner has not yet achieved a stable grasp of the knowledge element itself and its relationships. Therefore, the corresponding element mastery status and related application status should be lowered respectively.

[0124] When the answer to a verification item in the association application meets the preset pass conditions, the corresponding association application status is increased, and the teaching effectiveness of the teaching resource combination sequence for the association relationship is recorded. The teaching effectiveness record may include resource sequence identifier, target association relationship, verification result, and execution time.

[0125] When the verification task only includes knowledge element verification items, the mastery status of the corresponding element is updated according to the answer to the knowledge element verification items, while the associated application status remains unchanged; when the verification task only includes associated application verification items, the associated application status is updated according to the answer to the associated application verification items, while the mastery status of the relevant knowledge elements remains in the pre-diagnosis state until the knowledge element verification result is obtained.

[0126] The status update magnitude or status level transition rules can be pre-configured before the system runs. For example, a status level transition table can be used to determine the status changes when verification passes and fails. The updated status generated after the current verification is completed is written to the learning record database and used for subsequent question recommendations, subsequent error diagnosis, or subsequent teaching resource allocation.

[0127] The target error set and teaching resource combination sequence for this diagnosis are both generated based on the learning state before the update. The updated state obtained after verification is only used for subsequent learning cycles and is not used as input for the determination process of the target error set, thereby avoiding the formation of a logical loop of state updates.

[0128] In one possible implementation, knowledge-related data is stored using a directed graph. Each knowledge node has a unique knowledge element identifier, and each directed edge has a starting node identifier, a target node identifier, a relationship type, and a relationship direction. The learner's element mastery status is stored in association with the knowledge element identifier, and the application status is stored in association with the directed edge identifier.

[0129] For example, in the teaching scenario of "solving fractional equations", "basic properties of fractions", "determining the least common denominator", "eliminating denominators from equations", and "solving linear equations in one variable" can be set as multiple knowledge nodes, and the prerequisite relationships, representation transformation relationships, and reasoning succession relationships between them can be set as directed association edges. Different learners may have the same knowledge association structure, but they may have different levels of mastery and application of the elements.

[0130] In another possible implementation, each problem-solving step in the target problem-solving knowledge path can also be configured with allowed equivalent operations. If a learner adopts a processing method that differs from the standard problem-solving steps but yields an equivalent result, it is judged as a match based on the equivalent operation configuration, avoiding misjudging reasonable alternative solutions as abnormal problem-solving positions.

[0131] For example, a standard problem-solving procedure requires moving terms first and then merging like terms, while learners merge like terms first and then move terms. If both operations satisfy the corresponding input and output conditions, the learner can continue to execute the next matching step according to the preset equivalent operation relationship.

[0132] In another possible implementation, historical correct answers and historical incorrect answers are stored chronologically. When screening for initial error-causing factors, priority can be given to historical records that belong to the same teaching stage as the target incorrect question, have similar difficulty levels, and are from a more recent time period. This reduces the interference of earlier learning records or questions with significantly different difficulty levels on the current diagnosis.

[0133] In one possible implementation, the predicted response results are represented by multiple levels, such as low probability of correctness, medium probability of correctness, and high probability of correctness. Both the baseline predicted response results and the predicted response results after hypothetical state adjustment use the same level system. If adjusting a candidate error-causing factor increases the predicted level from low probability of correctness to high probability of correctness, then that candidate error-causing factor meets a preset improvement condition.

[0134] In another possible implementation, the predicted answer value is a value within a preset range. The preset improvement criteria can be configured based on historical diagnostic records and determined before the current diagnostic begins. Different subjects or different question types can use different preset improvement criteria.

[0135] In one possible implementation, candidate teaching resources come from a school-based resource repository and a cloud-based resource repository. The teaching resource matching module can first select resources from the school-based resource repository based on the target learner's school, textbook version, and teaching progress, and then supplement resources from the cloud-based resource repository when school-based resources are insufficient.

[0136] In another possible implementation, a consistency check is performed on the resource versions. When multiple teaching resources correspond to different textbook versions, and the definitions of knowledge elements or the order of problem-solving are inconsistent in the different versions, the multiple teaching resources are not grouped into the same resource sequence.

[0137] In another possible implementation, when the relationships in the knowledge-related data are modified, the teaching resource coverage information and existing resource combination records that depend on those relationships are marked. New error diagnosis uses the modified relationships, while completed historical diagnosis results retain their original versions for easy tracking of diagnostic basis.

[0138] In one possible implementation, instructional effectiveness records are used for subsequent resource selection. When multiple candidate resources are similar in terms of resource difficulty, expected learning time, and content repetition, resources with more successful validation records for the same relationship can be prioritized. The instructional effectiveness records generated in the current validation are only used for subsequent resource allocation and do not affect the selection of currently completed resource sequences.

[0139] The following example, using the solution of product-form equations, illustrates the complete execution process of the method described in this application. This example is only used to explain the data flow and processing relationships between the steps and does not constitute a limitation on the type of question or the knowledge content.

[0140] The teaching server pre-stores the knowledge elements "property of zero product" and "solution of linear equations in one variable," as well as the conditional triggering relationships and reasoning connections between them. The teaching resource database stores resources explaining the property of zero product, demonstration resources for zero product conversion, comparative analysis resources on products of zero and sums of zero, and transfer exercises on product equations.

[0141] Target learners complete the "Solve" section of the target error questions. At that time, the system directly performed incorrect operations on the two factors, failing to perform the conversion from a product equal to zero to two linear equations in one variable. The student terminal then sent the intermediate input results, modification operations, and final answer to the teaching server.

[0142] The problem-solving path processing module generates the target problem-solving knowledge path, and the answer process matching module matches the learner's answer process with the order of the target problem-solving knowledge path, and identifies the step of "calling the zero product property for formal transformation" as an abnormal problem-solving location.

[0143] The error analysis module traces back along the target problem-solving knowledge path to identify the zero product property, the solution method for linear equations in one variable, and the relationship between the two as the initial error-causing factors. Historical learning records show that the target learner was previously able to correctly answer direct judgment questions about the zero product property and correctly solve linear equations in one variable, but failed to correctly complete the form transformation in multiple product-form equations.

[0144] After filtering based on historical records, the "association between a product equal to zero and the transformation into two linear equations in one variable" was retained as the primary candidate error-causing factor. The error cause analysis module determines the baseline answer prediction value based on the original learning state.

[0145] The error analysis module improves the mastery status of elements with zero-product properties in the temporary state copy and redetermines the predicted value of the first answer result; after restoring the original state copy, it improves the mastery status of elements with linear equation solving methods and determines the predicted value of the second answer result; after restoring the original state copy again, it improves the application status of the aforementioned relationships and determines the predicted value of the third answer result.

[0146] If the improvement of the third response prediction value relative to the baseline response prediction value meets the preset improvement condition, while the improvement of the first response prediction value and the second response prediction value does not meet the preset improvement condition, then the correlation is determined as the target error cause.

[0147] The teaching resource assembly module selects zero-product transformation demonstration resources, comparison and analysis resources, and transfer practice resources based on the correlation coverage information, and generates a teaching resource assembly sequence in the order of "association rule description - applicable condition analysis - independent transfer practice".

[0148] The student terminal outputs the teaching resources sequentially. After the target learner completes the learning of the resources, the teaching server provides a related application verification item whose numerical value and expression are different from the target incorrect question, but still requires the use of the zero product property to complete the equation transformation.

[0149] If the learner correctly completes the verification item related to the association, the status update module increases the corresponding association status and records the teaching effectiveness of the teaching resource combination sequence for that association. If the learner still fails to complete the verification item correctly, the association status is maintained or decreased, and a lower-difficulty or more detailed association bridging resource can be selected in subsequent learning cycles.

[0150] Therefore, this application can independently diagnose and update the application status of the relationships between knowledge elements while maintaining the status of verified knowledge elements, reducing the possibility of inaccurate adjustments to multiple knowledge elements due to errors in comprehensive questions.

[0151] This application also provides a teaching resource allocation and error correction device based on knowledge point association. The device may include a data acquisition module, a path matching module, an error cause analysis module, a resource allocation module, and a verification and update module.

[0152] The data acquisition module is used to acquire knowledge-related data, teaching resource data, and learning records of target learners; the path matching module is used to generate target problem-solving knowledge paths based on the problem-solving information of target incorrect questions and determine the abnormal problem-solving locations; the error cause analysis module is used to identify candidate error-causing factors, perform hypothetical state adjustments, and determine the target error cause set.

[0153] The resource allocation module is used to determine the target teaching resources that cover the target error set and generate a teaching resource allocation sequence; the verification and update module is used to provide verification tasks to the target learners and update the corresponding status according to the verification object and the answer result of the verification task.

[0154] This application also provides an electronic device, which includes a processor, a memory, and a computer program stored in the memory and capable of running on the processor. When the processor executes the computer program, it implements the method steps described in any of the above embodiments.

[0155] The processor may be a central processing unit, a microprocessor, or other processing unit with data processing capabilities. The memory may include volatile memory and / or non-volatile memory, used to store knowledge-related data, teaching resource data, learning records, and computer programs.

[0156] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the method described in any of the above embodiments.

[0157] The above description is merely an exemplary embodiment of this application and is not intended to limit the scope of protection of this application. For those skilled in the art, various modifications and improvements can be made without departing from the technical concept of this application, and all such modifications and improvements should fall within the scope of protection of this application.

Claims

1. A method for combining teaching resources and reviewing incorrect answers based on knowledge point associations, characterized in that, include: Acquire knowledge-related data, teaching resource data, and learning records of the target learner. The knowledge-related data includes multiple knowledge elements and the relationships between the knowledge elements. The learning records include the element mastery status corresponding to the knowledge elements and the related application status corresponding to the relationships. The element mastery status and the related application status constitute the learning status of the target learner. Based on the solution information of the target incorrect questions, a target solution knowledge path is generated, and the answer process of the target learner for the target incorrect questions is matched with the target solution knowledge path to determine the abnormal solution location; Based on the abnormal problem-solving location and the learning record, candidate error-causing factors are determined from the target problem-solving knowledge path. Hypothetical state adjustments are made to each candidate error-causing factor. Based on the learning state after the hypothetical state adjustment, the predicted value of the target learner's answer to the target wrong question is determined. Based on the predicted value of the answer corresponding to each candidate error-causing factor, the target error cause set is determined. Based on the target error set and the correspondence between teaching resources and knowledge elements and / or relationships in the teaching resource data, target teaching resources covering the target error set are determined from the teaching resource data. Based on the sequential dependency between knowledge elements and / or relationships corresponding to each target teaching resource, the presentation order of each target teaching resource is determined, and a teaching resource combination sequence is generated. The teaching resource combination sequence is provided to the target learner in sequence, and after the teaching resource combination sequence is completed, a verification task corresponding to the target error set is provided to the target learner; based on the verification object corresponding to the verification task and the target learner's answer to the verification task, the corresponding element mastery status and / or associated application status are updated.

2. The method according to claim 1, characterized in that, The knowledge-related data is represented by a directed association structure, the knowledge elements constitute knowledge nodes in the directed association structure, and the association relationships constitute directed association edges connecting different knowledge nodes. The relationships include at least one of the following: prerequisite relationship, condition triggering relationship, representation transformation relationship, reasoning inheritance relationship, and joint application relationship; For each knowledge node, a corresponding element mastery state is set, and for each directed edge, a corresponding association application state is set.

3. The method according to claim 1, characterized in that, The process of generating the target problem-solving knowledge path based on the problem-solving information of the target incorrect questions includes: Obtain the standard solution process corresponding to the target incorrect question, and divide the standard solution process into multiple solution steps arranged in the solution order; Determine the input state, problem-solving operation, and output state corresponding to each of the problem-solving steps, and determine the knowledge elements and / or relationships invoked when executing each of the problem-solving operations; According to the execution order of each problem-solving step, the problem-solving steps and the knowledge elements and / or relationships called by each problem-solving step are connected to generate the target problem-solving knowledge path.

4. The method according to claim 3, characterized in that, The step of matching the target learner's answer process for the target incorrect questions with the target problem-solving knowledge path to determine the abnormal problem-solving location includes: Multiple learner problem-solving steps are extracted from the question-answering process. The learner problem-solving steps include intermediate answer results and answering operations used to obtain the intermediate answer results. Based on the consistency of input state, equivalence of answering operation, and consistency of output state between each learner's problem-solving step and each problem-solving step in the target problem-solving knowledge path, the learner's problem-solving steps are sequentially matched with each problem-solving step; The position of the first learner's problem-solving step that does not meet the preset matching conditions in the target problem-solving knowledge path is determined as the abnormal problem-solving position.

5. The method according to claim 4, characterized in that, The step of determining candidate error-causing factors from the target problem-solving knowledge path based on the abnormal problem-solving location and the learning record includes: Starting from the abnormal solution location, backtrack along the reverse execution direction of the target solution knowledge path to determine the solution steps corresponding to the abnormal solution location, as well as the knowledge elements and relationships upstream of the solution steps; The knowledge elements and relationships obtained through backtracking are identified as the initial error-causing factors; Extract historical correct answers and historical incorrect answers corresponding to each of the initial error-causing factors from the learning records. Filter each of the initial error-causing factors based on the historical correct answers and historical incorrect answers to obtain the candidate error-causing factors.

6. The method according to claim 5, characterized in that, The step of performing hypothetical state adjustments on each of the candidate error-causing factors and determining the target error cause set based on the predicted answer values ​​corresponding to each candidate error-causing factor includes: Based on the learning state before hypothetical state adjustment, determine the baseline predicted value of the target learner's answer to the target incorrect question; When the candidate error-causing factor is a knowledge element, the mastery status of the element corresponding to the candidate error-causing factor is temporarily adjusted to the preset mastery status, while keeping the mastery status and associated application status of other elements unchanged. When the candidate error-causing factors are related, the associated application state corresponding to the candidate error-causing factors is temporarily adjusted to the preset application state, while keeping other associated application states and element mastery states unchanged. Based on the learning state after each hypothetical state adjustment, the corresponding predicted value of the answer result is determined, and the target error cause set is determined from the candidate error-causing factors according to the change of each predicted value of the answer result relative to the baseline predicted value of the answer result.

7. The method according to claim 1, characterized in that, The teaching resource data also includes knowledge element coverage information, correlation coverage information, resource function type, resource difficulty and expected learning time for each teaching resource; The step of determining the target teaching resources covering the target error set based on the target error set includes: Based on the knowledge element coverage information and the relationship coverage information, candidate teaching resources that can cover each target error in the target error set are determined; The target teaching resource is determined from the candidate teaching resources based on the difficulty of the resource, the expected learning time, and the degree of content overlap between different candidate teaching resources. Wherein, when the target error set includes correlations, the target teaching resources include at least one correlation bridging resource for explaining the correlations.

8. The method according to claim 7, characterized in that, The generated teaching resource combination sequence includes: Based on the prerequisite relationships between the knowledge elements and related relationships covered by each of the target teaching resources, the knowledge dependency relationships between the target teaching resources are determined. Based on the resource function type of each of the target teaching resources, determine the functional connection relationship between each of the target teaching resources; Based on the knowledge dependency relationship and the functional connection relationship, the target teaching resources are sorted to generate a teaching resource combination sequence that includes at least two of the following: knowledge explanation resources, association bridging resources, comparison and analysis resources, and transfer practice resources.

9. The method according to claim 1, characterized in that, The verification task corresponding to the target error set includes at least one of knowledge element verification items and related application verification items; The knowledge element verification item is used to verify the target learner's mastery of the knowledge elements in the target error set; The knowledge elements and relationships invoked by the verification item are the same as those invoked by the target incorrect question, and the verification item differs from the target incorrect question in at least one of the following: question object, question value, language expression, and application scenario.

10. The method according to claim 9, characterized in that, The step of updating the corresponding element mastery status and / or associated application status based on the verification object corresponding to the verification task and the target learner's response to the verification task includes: When the answer to the knowledge element verification item meets the preset pass condition, and the answer to the related application verification item does not meet the preset pass condition, the element mastery status corresponding to the knowledge element verification item is maintained, and the related application status corresponding to the related application verification item is reduced. When the answer results of the knowledge element verification item and the related application verification item do not meet the preset pass conditions, the corresponding element mastery status and related application status are reduced respectively. When the answer to the related application verification item meets the preset pass condition, the related application status corresponding to the related application verification item is improved, and the teaching effectiveness of the teaching resource combination sequence on the related relationship is recorded.

Images