Teaching optimization method, system and equipment based on knowledge space and medium

Abstract

The invention discloses a teaching optimization method, system and device based on knowledge space and a medium. The method specifically comprises the steps that a secondary node platform is analyzed based on national industrial internet identification, official teaching materials are analyzed to generate knowledge point entities with identification codes, and a relation network is established; based on the relational network, collecting a multi-platform question bank problem solving step, performing authentication in combination with an identification code matching mechanism, synchronizing to a basic knowledge space after authentication, marking as an extended knowledge point library, and calculating a difficulty threshold value of a knowledge point association combination in combination with wrong question statistics; and on the basis of the relational network, cross-platform problem solving steps of the learner are captured in real time, authentication is carried out in combination with an identification code matching mechanism, if authentication is not passed, a preposed knowledge point unmastered state is marked, and a learner problem solving knowledge space with a state mark is generated. According to the invention, comprehensive analysis of the knowledge point mastering condition of the learner, difficult point positioning and personalized teaching optimization suggestion generation are realized, and the teaching quality and the learning effect are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and particularly relates to a teaching optimization method and system based on a knowledge space, a device and a medium. BACKGROUND

[0002] In the field of education today, teaching quality analysis is of great significance for improving teaching effectiveness, optimizing teaching strategies, and helping students better master knowledge. At present, most teaching quality analysis mainly relies on learning behavior data and problem-solving data of learners. However, this analysis method has many limitations, resulting in inaccurate analysis of learners, and thus affecting the recommendation accuracy of learning programs and test types.

[0003] On the one hand, there is an incomplete problem in obtaining learning behavior data of learners. Learners may complete learning of a knowledge point through offline or other channels or methods, and these learning behavior data cannot be obtained by existing analysis systems. For example, students participate in after-school tutoring, read related books, or conduct group discussions, and the knowledge and experience accumulated through these learning activities cannot be included in the existing teaching quality analysis system. This makes the basic analysis of learners' learning not comprehensive enough, and the overall learning situation of learners cannot be accurately mastered, which ultimately leads to biased analysis results of learners and cannot truly reflect the knowledge mastery and ability level of learners.

[0004] On the other hand, many current technologies are based on whether a learner's answer to a certain type of question is correct or not to make a judgment, and then analyze the learner's mastery of the knowledge point related to the question type. This analysis and judgment method is too general and cannot accurately pinpoint the learner's knowledge point mastery and application defects. For example, for the same type of question, the learner answers correctly in question A and incorrectly in question B, and from the accuracy rate of the question type, it cannot be determined that the learner has a problem in which specific knowledge point, it may be in the aspect of text understanding, it may be in the aspect of knowledge point application, or it may be caused by other related factors. Since the knowledge point mastery and application defects cannot be accurately pinpointed, the recommended learning programs or test types are also not accurate, which cannot meet the individualized learning needs of learners and is not conducive to better improving the knowledge and ability of learners.

[0005] In addition, current technologies are based on obtaining learning records of learners on platforms to analyze the situation of students on the learning path, but the data of each platform is saved independently, and only provides extracted knowledge points and word segmentation information. This data island phenomenon makes it impossible to comprehensively and deeply analyze the learning process and knowledge mastery of learners. For example, different platforms may have differences in the division and expression of knowledge points, making data integration difficult and making it difficult to form a unified and complete knowledge portrait of learners. At the same time, only providing extracted knowledge points and word segmentation information, lacking analysis of detailed information such as problem solving steps, cannot deeply understand the thinking process and knowledge application of learners in the problem solving process, further limiting the accuracy and effectiveness of teaching quality analysis. SUMMARY

[0006] The purpose of the present application is to provide a knowledge space-based teaching optimization method, system, device and medium, which realizes comprehensive analysis of the knowledge point mastery of learners, difficulty positioning and personalized teaching optimization suggestion generation by constructing a knowledge space system, and improves the teaching quality and learning effect, to solve at least one of the above technical problems.

[0007] In a first aspect, the present application provides a knowledge space-based teaching optimization method, which specifically comprises:

[0008] Based on the national industrial internet identifier resolution secondary node platform, the official textbook is parsed to generate knowledge point entities with identifier codes, and a relationship network containing sequential, parallel and comprehensive types is established;

[0009] Based on the relationship network, the problem solving steps of the multi-platform question bank are collected, authenticated by combining the identifier code matching mechanism, synchronized to the basic knowledge space after authentication and marked as an extended knowledge point library, and the difficulty threshold of the knowledge point association combination is calculated by combining the mistake statistics;

[0010] Based on the relationship network, the learner's cross-platform problem solving steps are captured in real time, authenticated by combining the identifier code matching mechanism, and if the authentication fails, the pre-knowledge points are marked as unmastered state, and a learner problem solving knowledge space with state markers is generated;

[0011] For answer questions, the learner and the standard answer path are mapped by a path comparator, the feature state value is calculated based on a multi-dimensional analysis model, and the knowledge point mastery rate is calculated;

[0012] For selection and judgment questions, the same type of answer questions are matched by a question type associator, and the error type is inferred according to the feature state of the associated answer questions;

[0013] The high coincidence type is clustered based on head and tail information of a problem solving path, class data is projected to a stage knowledge space, a multi-dimensional heat map is rendered according to a knowledge point mastery rate to visualize knowledge defect distribution, and a knowledge point mastery path optimization suggestion is generated by mining a weight relationship between the knowledge point mastery degree and a problem solving feature through an Apriori algorithm.

[0014] In a second aspect, the present application provides a knowledge space-based teaching optimization system, which specifically comprises:

[0015] A first teaching optimization module is configured to parse a secondary node platform based on a national industrial internet identifier, parse official teaching materials to generate knowledge point entities with an identifier code, and establish a relationship network including a sequential type, a parallel type and a comprehensive type.

[0016] A second teaching optimization module is configured to collect problem solving steps of a multi-platform question bank based on the relationship network, authenticate through a matching mechanism based on the identifier code, synchronize to a basic knowledge space and mark as an extended knowledge point library after authentication, and calculate a difficulty threshold of a knowledge point associated combination in combination with a mistake question statistics.

[0017] A third teaching optimization module is configured to capture learning steps across platforms in real time based on the relationship network, authenticate through a matching mechanism based on the identifier code, mark as a pre-knowledge point unmastered state if the authentication fails, and generate a learner problem solving knowledge space with a state mark.

[0018] A fourth teaching optimization module is configured to map a learner and a standard answer path through a path comparator for a solved question, calculate a feature state value based on a multi-dimensional analysis model and calculate a knowledge point mastery rate.

[0019] A fifth teaching optimization module is configured to match a same type solved question through a question type associator for a selection and judgment question, and then deduce a mistake type according to a feature state of the associated solved question.

[0020] A sixth teaching optimization module is configured to cluster a high coincidence type based on head and tail information of a problem solving path, project class data to a stage knowledge space, render a multi-dimensional heat map according to a knowledge point mastery rate to visualize knowledge defect distribution, and generate a knowledge point mastery path optimization suggestion by mining a weight relationship between the knowledge point mastery degree and a problem solving feature through an Apriori algorithm.

[0021] In a third aspect, the present application provides a computer device, which comprises a memory, a processor and a computer program stored in the memory, and when the computer program is executed on the processor, a knowledge space-based teaching optimization method as described in any of the above methods is implemented.

[0022] In a fourth aspect, the present application provides a computer readable storage medium, having stored thereon a computer program, which, when run by a processor, implements the knowledge space-based teaching optimization method according to any one of the above methods.

[0023] Compared with the prior art, the present application has at least one of the following technical effects:

[0024] 1. The present application realizes comprehensive analysis of the mastery of knowledge points by learners, difficulty positioning and personalized teaching optimization suggestion generation by constructing a knowledge space system, thereby improving teaching quality and learning effect.

[0025] 2. The present application uses a national industrial internet identifier resolution secondary node platform to accurately resolve textbook-generated knowledge point entities with identifier codes and construct a multi-type relationship network, thereby ensuring the integrity and scalability of the knowledge system and providing a solid foundation for subsequent teaching optimization.

[0026] 3. The present application collects problem-solving steps from multiple platforms and authenticates them to the basic knowledge space, combines wrong question statistics to calculate knowledge point association combination difficulty thresholds, and effectively identifies teaching difficulties, thereby providing data support for targeted teaching.

[0027] 4. The present application captures learners' problem-solving steps in real time and authenticates them, marks the state of not mastering the prerequisite knowledge points to generate a knowledge space with state markers, accurately locates the weak links in learners' knowledge, and facilitates timely intervention and guidance.

[0028] 5. The present application calculates characteristic state values and statistics knowledge point mastery rates through path comparison and multi-dimensional analysis model for answer questions, thereby comprehensively evaluating learners' mastery of knowledge points for answer questions and providing a basis for personalized learning.

[0029] 6. The present application accurately locates the reasons for learners' mistakes in multiple-choice questions by matching the same type of answer questions and deducing the types of mistakes, generates targeted learning remediation instructions, and improves learning efficiency.

[0030] 7. The present application clusters question types based on the head and tail information of the problem-solving path, projects class data to the stage knowledge space and generates a multi-dimensional heat map to visualize the distribution of knowledge defects, and mines the relationship between knowledge point mastery and problem-solving feature weights, thereby providing intuitive and scientific suggestions for class teaching optimization.

[0031] 8. Unlike existing techniques that generally judge knowledge mastery based on question type and accuracy, this invention analyzes learners' mastery of each knowledge point. By analyzing the problem-solving steps in detail, it can pinpoint where learners have gaps in mastery and application, such as text comprehension issues and lack of proficiency in knowledge point application. This allows learners to receive more targeted learning recommendations and test question type assessments, helping them to better improve their knowledge and abilities.

[0032] 9. This invention establishes a foundational knowledge space, a knowledge application space, and a learner's problem-solving knowledge space. It then projects the school's phased knowledge space, generated based on teaching progress and exam syllabus requirements, onto each student's problem-solving knowledge space. This allows for the integration of data from multiple platforms to create a more complete learner knowledge profile. Furthermore, analysis of detailed information, such as problem-solving steps, provides a deeper understanding of learners' thought processes and knowledge application, improving the accuracy and effectiveness of teaching quality analysis. This allows teachers to more clearly understand students' overall knowledge mastery, their various abilities, and their knowledge gaps, providing strong support for teaching decisions.

[0033] 10. Based on the precise identification of learners' knowledge mastery and application deficiencies, this invention can generate personalized learning plans and precise test questions for learners. By providing targeted recommendations for learning content and test questions, it can meet learners' personalized learning needs, improve learning efficiency and effectiveness, and help students better master knowledge and enhance their abilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0035] Figure 1 This is a flow chart of a teaching optimization method based on knowledge space provided by one embodiment of the present invention;

[0036] Figure 2 1 is a schematic structural diagram of a teaching optimization system based on a knowledge space provided by an embodiment of the present invention;

[0037] Figure 3 It is a structural diagram of a computer device provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0038] In the following description, for purposes of explanation and not limitation, specific details are set forth such as particular architectures, techniques, etc. in order to provide a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application can be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known methods, devices, circuits, and

[0039] It will be understood that the term "includes," "including," "has," "having," "comprises," "comprising," "contains" or "containing," when used in this specification and in the following claims, specifies the presence of the stated features, integers, steps, operations, elements, and / or components but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0040] It will also be understood that the term "and / or," when used in this specification and in the following claims, can connote any conjunctive or disjunctive sense, in any combination, and can include any of the possible combinations of the items linked by the term.

[0041] As used in this specification and in the claims, the term "if" can be construed to mean "when" or "upon" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrase "if it is determined" or "if [a described condition or event] is detected" can be construed to mean "upon determining" or "in response to determining" or "upon detecting [the described condition or event]" or "in response to detecting [the described condition or event]," depending on the context.

[0042] In addition, the terms "first," "second," "third," etc. are used herein only to describe different instances of an element, and do not imply relative importance of the elements.

[0043] The terms "one embodiment," "some embodiments," "an embodiment," "some embodiments," etc. as may be used herein, mean that the particular feature, structure, or characteristic following the term is included in at least one embodiment of the present application. Thus, use of the terms "in one embodiment" or "in some embodiments" or "in an embodiment" or "in some embodiments" or the like, in various places throughout this specification, does not necessarily refer to the same embodiment, although it can. Furthermore, the terms "first," "second," "third," etc. are used herein only to describe different instances of an element, and do not imply relative importance of the elements. The terms "including," "containing," "comprising," "having," and the like are used herein to mean "including but not limited to." The terms "coupled," "connected," and "in communication with" are used herein to express a relationship between or among two or more elements, and are not necessarily limited to a direct connection between the elements.

[0044] In the embodiments of the present application, the execution subject of the flow includes a terminal device. The terminal device includes but is not limited to a server, a computer, a smart phone, a tablet computer and other devices capable of executing the method disclosed in the present application. Figure 1 A flowchart of the knowledge space-based teaching optimization method disclosed in an embodiment of the present application is shown, and is described in detail as follows:

[0045] In S101, the official textbook is parsed to generate knowledge point entities with identification codes based on the national industrial internet identification resolution secondary node platform, and a relationship network including sequential type, parallel type and comprehensive type is established.

[0046] In the embodiments, the construction of the real-time knowledge space includes the construction of the basic knowledge space, the knowledge application space and the learner's problem-solving knowledge space. The corresponding knowledge space is uploaded to the national industrial internet identification resolution secondary node platform in real time, and is used for information retrieval and application of the cross-platform teaching platform.

[0047] For a learner's problem-solving knowledge space, the learner's knowledge space can be retrieved on different learning platforms, and the knowledge resources and question bank of the platform can be analyzed and pushed.

[0048] For a school or a teacher, the knowledge space information of the related students can also be obtained under the condition of authentication and authorization, and the students can be comprehensively analyzed.

[0049] The construction of the basic knowledge space is based on the national textbook and the related official teaching course to generate a relatively authoritative basic knowledge space. A unique identification code is generated for each knowledge point in a certain knowledge field based on a certain subject, and the relationship between the knowledge points is obtained, including sequential type (including causal relationship, progressive relationship and hierarchical relationship), parallel type (knowledge points exist in parallel, independent of each other, and have no necessary connection with each other) and comprehensive type (knowledge points are related to each other and are applied comprehensively).

[0050] Based on the sequential relationship, a basic "knowledge point-relation-knowledge point" triple is established, and a unique identification code is generated for each "entity-relation-entity" triple.

[0051] Based on the comprehensive relationship, a "multi-entity-multi-relation-multi-entity" relationship combination is established to generate a unique identification code.

[0052] The teaching courseware and the course content are matched with the entity identification code corresponding to the knowledge point, the sequential relationship combination identification code and the comprehensive relationship combination identification code, and the knowledge point is marked as a basic knowledge point library.

[0053] S102, based on the relationship network, collect multi-platform question bank problem solving steps, combine the identification code matching mechanism for authentication, after authentication, synchronize to the basic knowledge space and mark as an extension knowledge point library, and at the same time, combine the wrong question statistics to calculate the difficulty threshold of the knowledge point association combination.

[0054] In this embodiment, the question bank of the school or each teaching platform and the corresponding problem solving steps are obtained, the knowledge points and knowledge point reasoning relationship combinations applied in each step of the problem solving steps are split, and the knowledge points and knowledge point reasoning relationship combinations applied in the problem solving are matched to the entity identification code corresponding to the knowledge points and the knowledge point relationship combinations generated in step S101.

[0055] If the knowledge points or knowledge point relationship combinations applied in the problem solving cannot be matched to the identification code, the knowledge points or knowledge point relationship combinations are pushed to the expert review system of the secondary node for authentication by experts. If the review is passed, the experts define the discipline and classification, generate a unique identification code for the knowledge points in the knowledge points or knowledge point relationship combinations, and generate an identification code for the knowledge point relationship combinations. The problem solving steps are associated with the knowledge points and knowledge point relationship combinations. At the same time, the corresponding knowledge points and knowledge point relationship combinations are also synchronized to the public basic knowledge space library and marked as an extension of the first knowledge point library in the basic knowledge space library. If the review is not passed, the validity of the current problem solving step is marked as to be determined.

[0056] If the knowledge points or knowledge point relationship combinations applied in the problem solving can be matched to the identification code, the problem solving steps are associated with the knowledge points and knowledge point relationship combinations.

[0057] As described above, the final problem solving knowledge point association path of each question is formed. If a question has multiple problem solving steps, multiple problem solving knowledge point association paths are formed according to the above method.

[0058] The problem solving steps of the learning subjects who answered the questions incorrectly are obtained, the knowledge point association path of the learning subjects is obtained, the knowledge points and knowledge association combinations of each step in the knowledge point association path of the question and the learning subject's answer are compared, the number of learning subjects' answers that do not match each step Gn and the number of learning subjects' answers that match each step Hn are counted, the accuracy rate Ln of each step is calculated, and when Ln is less than a threshold value, the corresponding knowledge point and knowledge association combination of this step is recorded as a difficult point. Further, the difficult point of the knowledge point and knowledge association combination corresponding to the problem solving of each question is identified.

[0059] S103, based on the relationship network, real-time capture of learning subjects' cross-platform problem solving steps, combined with identification code matching mechanism for authentication, if not authenticated, mark as pre-knowledge point not mastered state, generate a learning subject problem solving knowledge space with state mark.

[0060] In this embodiment, the problem solving steps of the learner's answers to the problems in a school or teaching platform are obtained, and the knowledge points and knowledge point reasoning relationship combinations applied in each problem solving step are split and combined. The knowledge points and knowledge point reasoning relationship combinations applied in the problem solving are matched to the entity identification code corresponding to the knowledge points and the knowledge point relationship combination generated in step S101.

[0061] If the knowledge points or knowledge point relationship combinations applied in the problem solving cannot be matched to the identification code, the knowledge points or knowledge point relationship combinations are pushed to the expert review system of the secondary node for authentication by experts. If the review is passed, the experts define the discipline and classification, generate a unique identification code for the knowledge points in the knowledge points or knowledge point relationship combinations, and generate an identification code for the knowledge point relationship combination. The problem solving steps are associated with the knowledge points and knowledge point relationship combinations. At the same time, the corresponding knowledge points and knowledge point relationship combinations are also synchronized to the public basic knowledge space library and marked as an extended second knowledge point library in the basic knowledge space library. If the review is not passed, the current problem solving step of the learner is marked as incorrect, and the learner is marked as not having mastered the pre-knowledge points corresponding to the current problem solving step.

[0062] If the knowledge points or knowledge point relationship combinations applied in the problem solving can be matched to the identification code, the problem solving steps are associated with the knowledge points and knowledge point relationship combinations.

[0063] As described above, the final learner's problem solving knowledge point association path for all problem solving is formed.

[0064] As a brief summary of the above steps S101-S103, for example, the first step (1) sample data acquisition: obtaining part of the problem solving and problem solving step data of each chapter as sample problem solving data.

[0065] The second step (2) sample data splitting and marking: the problem solving steps in the sample problem solving data are split and marked as the smallest problem solving unit. The reasoning association knowledge relationship between each problem solving unit in the problem solving step is defined as unknown (such as theorem, law, and other knowledge), the tool method applied in each problem solving unit in the problem solving step is defined as unknown (such as method matching, Dijkstra algorithm, etc.), and the reasoning path of each problem solving unit in the problem solving step is defined as unknown (such as forward thinking, reverse thinking, divide and conquer, etc.).

[0066] Third step (3) sample data labeling: the expert knowledge management system selects the knowledge points in the basic knowledge space as the corresponding knowledge points of the minimum problem solving unit in the sample problem solving step splitting in (2); selects the methods, algorithms, etc. in the basic knowledge space as the tool methods of the minimum problem solving unit in the sample problem solving step splitting in (2); selects the theorems, laws, etc. in the basic knowledge space as the inference association between the minimum problem solving units in (2); selects the thinking strategies (forward thinking, reverse thinking, divide and conquer, etc.) in the basic knowledge space as the inference path between the minimum problem solving units or between multiple minimum problem solving units in (2).

[0067] Fourth step (4) establish a word segmentation library: for example, establish the logical words of the inference application between each problem solving unit in the problem solving, that is, the problem solving steps such as “because...so...”, “or”, “and” and the like.

[0068] Fifth step (5) model establishment: the labeled sample data is used as the training set, the verification set and the test set, and the extraction model of each dimension information of the problem solving step of each question is established, including the corresponding knowledge points of each problem solving unit and the predicted score of the corresponding knowledge points of the problem solving unit, the corresponding tool methods of each problem solving unit and the predicted score of the corresponding tool methods of the problem solving unit, the inference association between the problem solving units and the predicted score of the inference association, the inference path between multiple problem solving units and the predicted score of the inference path.

[0069] The sixth step (6) knowledge application space library / learner problem solving knowledge space: based on the word segmentation library, the system automatically splits and marks each minimum problem solving unit in the problem solving step data of the corresponding standard answer or the problem solving data of the learner's problem solving, and automatically marks the corresponding logical word segmentation. The problem solving step data marked by the system is imported into the model to extract the extraction information data of the corresponding problem solving step, including the knowledge points corresponding to each problem solving unit and the predicted scores of the knowledge points corresponding to the problem solving unit, the tool methods corresponding to each problem solving unit and the predicted scores of the tool methods corresponding to the problem solving unit, the reasoning associations between the problem solving units and the predicted scores of the reasoning associations, the reasoning paths between multiple problem solving units and the predicted scores of the reasoning paths. For the predicted scores reaching the threshold value, it is defined as valid information, and for the predicted score value not reaching the threshold value, it is defined as to-be-audited information. At the same time, one problem solving unit or multiple problem solving units and logical word segmentation that do not mention knowledge points, tool methods, reasoning associations, reasoning paths are also defined as to-be-audited information. The to-be-audited information is audited or corrected by the expert auditing system. For the knowledge points, tool methods, reasoning associations, reasoning paths and other information that exceed the basic knowledge space, the expert can define according to the specific situation during the audit, and synchronize to the basic knowledge space to expand the related knowledge of the basic knowledge space. If this audit information comes from the problem solving step of the platform standard answer, the expanded knowledge points, tool methods, reasoning associations, reasoning paths are marked as extending the first knowledge point library; if this audit information comes from the problem solving step of the platform learner problem solving, the expanded knowledge points, tool methods, reasoning associations, reasoning paths are marked as extending the second knowledge point library; as above, the to-be-audited information is converted into the knowledge points, tool methods, reasoning associations, reasoning path data corresponding to the problem solving unit that passes the audit. In this way, the standard answer and the problem solving step data of the learner's problem solving of the question bank are mapped one by one into the basic knowledge space. The problem solving step data is standardized, and the problem solving path data of each question standard answer and the learner's problem solving is generated.

[0070] S104, for answering questions, mapping the learner and the standard answer path through the path comparator, calculating the characteristic state value based on the multi-dimensional analysis model and calculating the knowledge point mastery rate.

[0071] In this embodiment, for answering questions, for the questions that have marked that the learner does not master the pre-knowledge points corresponding to the current problem solving step, the learner is marked based on the marked pre-knowledge points that the learner does not master. Define the characteristic state of the learner for this knowledge point as M0 (not mastered), and push the related knowledge point course or remind to practice and verify.

[0072] For questions that have not marked that the learner does not master the pre-knowledge points corresponding to the current problem solving step:

[0073] (1) Match the knowledge point association path corresponding to the problem solving steps of the learner and the knowledge point association path of the question answer (if there are multiple knowledge point association paths, multiple can be compared), map the knowledge point association path of the question answer to the knowledge point association path corresponding to the problem solving steps of the learner, and then analyze:

[0074] a. The certain proportion (pre-set threshold value) of the knowledge point and knowledge association combination applied by the question answer can be found in the knowledge point and knowledge association combination in the knowledge point association path corresponding to the problem solving steps of the learner in order, the found knowledge point or knowledge association combination marks the learner as "mastered" based on this question, and marks the learner's analysis result in a dimension as 1 for this question, otherwise -1;

[0075] b. Compare the last step of the two knowledge point association paths. If the last step of the knowledge point is the same, the last step of the knowledge point or the knowledge association combination marks the learner as "mastered" based on this question, and marks the learner's analysis result in b dimension as 2 for this question, otherwise -2;

[0076] c. Determine whether the learner's problem solving final result is accurate. If accurate, mark the learner's analysis result in c dimension as 3 for this question, otherwise -3;

[0077] d. Whether the knowledge point and knowledge association combination applied in the problem solving knowledge point association path of the learner contains the extended second knowledge point library in the public basic knowledge space library. If the applied knowledge point and knowledge association combination contains the extended second knowledge point library in the public basic knowledge space library, mark the learner as "mastered" based on this question. If so, mark the learner's analysis result in d dimension as 4 for this question, otherwise -4;

[0078] (2) Analyze the characteristics of the learner for the judgment result in (1):

[0079] 1) If a = 1 and b = 2 and c = 3, analyze the learner's mastery of the knowledge points and knowledge associations involved in the question answer, define the characteristic state as M1, and no longer push related knowledge point courses or questions;

[0080] 2) If a = 1 and b = 2 and c = -3, analyze the learner's mastery of the knowledge points and knowledge associations involved in the question answer, but careless mistake, define the characteristic state as M2, can push a lot of related knowledge point pure calculation reminders, improve calculation ability;

[0081] 3) If a = -1 and c = 3 and d = 4, analyze the learner's mastery of the knowledge points and knowledge associations involved in the problem solving, and the knowledge expansion ability is strong, define the characteristic state as M3, push other answer questions of the same type to test whether the knowledge points are mastered firmly;

[0082] As described above, then the knowledge points of the learner based on each answer question are acquired whether they have been mastered. The number E1 of each knowledge point marked by the learner as having been mastered is counted in all the answers of the learner (including the number of knowledge points of the extended second knowledge point library in the public basic knowledge space library that have been mastered), the number E2 of each knowledge point applied by the learner is counted in all the answers of the learner, and the mastery of the learner based on each knowledge point E=E1 / E2 is calculated.

[0083] In S105, for the selected judgment question, the same type of answer question is matched through the question type correlator, and the error type is inferred according to the feature state of the associated answer question.

[0084] In this embodiment, for the selected judgment question, 1) the learner's answer results are compared with the correct results, for the questions answered correctly by the learner, there may be two cases of lucky guess and actual correct answer, and for the questions answered incorrectly by the learner, there may be two cases of careless mistake and actual mistake. 2) The knowledge point association path corresponding to the problem solving process of the selected judgment question of the learner is acquired, the last knowledge point and knowledge association combination in the knowledge point association path of the answer question are acquired, and the answer question whose last knowledge point and knowledge association combination are the same as those of the current selected judgment question is acquired. The knowledge point association path of the answer question is mapped to the knowledge point association path corresponding to the selected judgment question, and if a certain number of proportions (predefined threshold) of the knowledge points and knowledge association combinations applied by the answer question can be found in the knowledge points and knowledge association combinations in the knowledge point association path corresponding to the problem solving step of the current selected judgment question in order, it is determined that the current answer question and the current selected judgment question are of the same type.

[0085] The mastery of the learner for the above answer question of the same type as the selected judgment question is counted, the proportion of a certain number (the number can be predefined) of features of the latest problem solving of the learner in this type is analyzed, and the features with high proportion (M0, M1, M2, M3).

[0086] If the feature acquired is M0, but the learner's answer to the selected judgment question is correct, it is determined that the learner is lucky to guess the selected judgment question, the knowledge points are not mastered, and the same type of answer question practice is needed to verify and further explore the knowledge point defects of the learner.

[0087] If the feature acquired is M1 or M2, but the learner's answer to the selected judgment question is incorrect, it is determined that the learner is careless and makes a mistake in the selected judgment question, and a large number of reminders of related knowledge points for pure calculation can be pushed to improve the calculation ability.

[0088] If the learner's answer to the selection judgment question is wrong, it is determined that the learner's knowledge of this selection judgment question is not solid, and related knowledge point courses and the same type of answer practice are pushed to verify the learner's knowledge defects and further dig into the learner's knowledge defects.

[0089] In S106, the high coincidence question type is clustered based on the head and tail information of the problem solving path, the class data is projected to the stage knowledge space, the multi-dimensional heat map is rendered according to the knowledge point mastery rate, the knowledge defect distribution is visualized, and the Apriori algorithm is used to mine the weight relationship between the knowledge point mastery degree and the problem solving feature, and the knowledge point mastery path optimization suggestion is generated.

[0090] In this embodiment, the knowledge point association path of all question answers is obtained, the last knowledge point and knowledge association combination applied by the answer are obtained as the tail information, the first knowledge point and knowledge association combination applied by the answer are obtained as the head information, the questions are grouped based on the tail information and the head information, and the questions are grouped according to the knowledge point association path of each question.

[0091] For a learner, the problem solving features of the same group of questions are analyzed, and in addition to M2, there are other M0, M1, M3 features, then it can be analyzed that the learner's ability of text understanding or scene application is weak, we define it as M4 = "weak understanding ability based on this question type", and push the learning resources to improve the understanding ability.

[0092] For each type of question, analyze the core knowledge points associated with it: based on the knowledge point mastery data of learners at the same stage and the characteristics data of the learners, such as the characteristics data of the learners in the school's grade. Obtain each learner's mastery data E for each knowledge point for all questions in a group based on a certain group of questions, while also counting the characteristics data M reflected by the learner based on each question (such as M0, M1, M2, M3, M4). Align all question knowledge points (for example: question 1 involves knowledge points a, b, c, question 2 involves knowledge points b, c, d, here the involved knowledge points include the learner's application of the extended second knowledge point library in the public basic knowledge space library and are marked as mastered, after alignment, the knowledge points are a, b, c, d, and the matching alignment coefficients are question 1 alignment coefficient (1, 1, 1, 0), question 2 alignment coefficient (0, 1, 1, 1)), eliminate data with the same alignment coefficient and different characteristics data reflected by the learner based on each group of questions, and remove duplicate data with the same alignment coefficient and the same characteristics data reflected by the learner based on each group of questions, and then form a knowledge point corresponding to the identification code as a column, for each row of data, multiply the learner's mastery data E for each knowledge point by the alignment coefficient, which equals the characteristics data M reflected by the learner based on this question, and then construct a knowledge point association model, and mine the association rules of the knowledge points through the Apriori algorithm. Through association analysis, obtain the influence weight of the mastery degree value of each knowledge point on the characteristic value reflected by this question, such as the higher or lower the mastery degree of knowledge point n, the greater the influence on the final answer to this question M certain characteristic reaction; In addition, the influence association between knowledge points can also be analyzed through association analysis, for example, the higher or lower the knowledge point m, the higher or lower the mastery of knowledge points m1, m2, m3...; again, for example, the higher or lower the knowledge points x1, x2, x3... the higher or lower the mastery of knowledge point x. Then help learners quickly master core knowledge points, help learners change the reaction characteristics to M1 or M3 in the good direction when answering a certain type of question, and finally easily master the answer to each type of question. At the same time, help learners understand the promotion relationship between knowledge points, and quickly establish a knowledge mastery path.

[0093] The school or class teacher can obtain the mastery value E of each knowledge point of all students in the whole grade or class, and perform numerical classification on the mastery value of each knowledge point, for example, E0 is 0, E1 is (0, 0.2], E2 is (0.2, 0.4], E3 is (0.4, 0.6], E4 is (0.6, 0.8], and E5 is (0.8, 1], and the mastery of the students in the whole grade or class at each knowledge point or knowledge combination is statistically generated. Based on the teaching progress of the current grade or class, a part of the public basic knowledge space library generated based on the progress is obtained, the learner problem solving knowledge space and problem solving feature data of the corresponding learner are projected into the part of the public basic knowledge space library, and the overall feature of each knowledge point and knowledge combination is obtained, for example, for each knowledge point or knowledge combination, how many students are in E0, E1, E2, E3, E4, and E5, and the student situation in different dimensions is represented by different colors and corresponding color depths, so that the school and the teacher can more clearly understand the overall knowledge point mastery and application situation of the students, and targeted teaching and practice can be strengthened; similarly, the analysis of the solution of the students based on a certain type of question is based on the proportion of the reaction features M0, M1, M2, M3, and M4 of each type of question, and targeted teaching or test testing is performed.

[0094] Based on the above steps S101-S106, in some embodiments, by obtaining the problem solving steps (including the problem solving standard answer and the problem solving steps of the learners passed through the review) of each type of question in the question bank and the knowledge points and knowledge reasoning process information of the associated knowledge space, a knowledge point application library is generated. The extended knowledge points and knowledge point association path links in each type of question are added to the knowledge point application library of this type of question, and then the knowledge point application library of each type of question is generated. Based on the information of the knowledge point application library of each type of question, the number of source knowledge points and reasoning association relationships of each knowledge point in the previous step is counted, the total number of each application knowledge point in the question type is counted as a high-frequency application knowledge point, and the reasoning proportion of each previous knowledge point is counted based on this high-frequency application knowledge point.

[0095] In some embodiments, in the above step S101, the secondary node platform based on the national industrial internet identifier is used to analyze official textbooks to generate knowledge point entities with identifier codes, and to establish a relationship network including sequential, parallel and comprehensive types, which specifically includes:

[0096] The semantic segmentation module of the secondary node platform based on the national industrial internet identifier is used to analyze official textbooks to generate independent knowledge units, and to assign a globally unique identifier code to each knowledge unit;

[0097] Based on the logical structure of the teaching materials, the sequential relationship is identified by the causal chain analyzer, the parallel relationship is extracted by the semantic similarity matcher, and the comprehensive relationship is generated by the cross-chapter association miner to form a relationship network.

[0098] A triple code containing entity-relationship-entity is created for sequential and parallel relationships, and a multi-entity and multi-relationship combination code is generated for comprehensive relationships.

[0099] When new teaching resources are added, the incremental parser extracts new knowledge points or relationships, submits them to the expert review system of the secondary node platform, and after authentication, assigns an identification code and updates the relationship network.

[0100] In this embodiment, in the field of education, official teaching materials are important carriers of knowledge dissemination, but the knowledge points in the teaching materials are scattered and lack systematic association identification, which is not conducive to teacher teaching and student self-learning. This embodiment aims to use the national industrial internet identification analysis secondary node platform to analyze official teaching materials, generate knowledge point entities with identification codes, and establish a relationship network containing sequential, parallel, and comprehensive relationships to provide a foundation for teaching optimization.

[0101] Specifically, ensure that the national industrial internet identification analysis secondary node platform is running normally, and the platform has a semantic segmentation module that can perform semantic understanding and analysis on text. Official teaching materials in the form of electronic documents are input into the platform. The teaching materials can be in various formats such as PDF, Word, etc., and the platform should have the corresponding document parsing capability to convert the document content into processable text data. The semantic segmentation module of the platform is used to perform semantic segmentation on the teaching material text. This module uses natural language processing technology to identify semantic boundaries in the text and divides the teaching material content into independent knowledge units. For example, in a mathematics textbook, a theorem, a formula, or a concept can be considered as an independent knowledge unit. Each independent knowledge unit is assigned a globally unique identification code. The identification code can be in the form of numbers, letters, or combinations thereof to ensure uniqueness in the platform's knowledge system. The identification code can be generated by the platform's built-in identification code generation algorithm, which can generate a unique identification code based on the characteristics and generation order of the knowledge unit.

[0102] The causal chain analyzer identifies sequential relationships in the teaching materials. It analyzes logical sentences in the teaching materials, such as "because... so..." and "first... then...", to determine the order and causal relationships between knowledge points. For example, in explaining the proof process of a mathematical theorem, there is a strict order between the steps, and the causal chain analyzer can identify the sequential relationships between these steps. For the identified sequential relationships, create a triple code containing entity-relation-entity. The entity is the identification code of the knowledge unit, and the relationship is "sequence" or similar descriptive vocabulary. Through the triple code, the order relationship between knowledge points can be clearly represented.

[0103] The semantic similarity matcher extracts parallel relationships in the teaching materials. The semantic similarity matcher calculates the semantic similarity between different knowledge units, and when the similarity exceeds a certain threshold, it is considered that these knowledge units have a parallel relationship. For example, in Chinese teaching materials, multiple words or sentences that describe the same thing may have a parallel relationship. For the extracted parallel relationships, also create a triple code containing entity-relation-entity, and the relationship is described as "parallel".

[0104] The cross-chapter association miner is called to generate comprehensive relationships. The cross-chapter association miner analyzes the knowledge point associations between different chapters in the teaching materials and discovers relationships that span chapters and integrate multiple knowledge points. For example, in physics teaching materials, mechanical and electrical knowledge may be related in some comprehensive application problems. For the generated comprehensive relationships, generate a multi-entity and multi-relation combination identification code to accurately describe the complex relationships between multiple knowledge points.

[0105] Integrate the identified sequential relationships, parallel relationships, and comprehensive relationships to form a complete knowledge point relationship network. The relationship network is stored in the form of a graph structure, with knowledge units as nodes and relationships as edges. The attributes of the edges are the relationship types (sequence, parallel, or comprehensive). Store the knowledge point relationship network in the database of the national industrial internet identification analysis secondary node platform. The database should have efficient data storage and query capabilities to facilitate subsequent access and updates to the relationship network.

[0106] When there are new teaching resources (such as new textbook chapters, teaching aids, etc.), the incremental parser is triggered. The incremental parser can automatically detect the knowledge points and relationships in the newly added resources. The incremental parser parses the newly added teaching resources and extracts new knowledge points or relationships. For new knowledge points, assign them an identification code according to the method in step one; for new relationships, create corresponding identification codes according to their types (sequential, parallel, or comprehensive). Submit the extracted new knowledge points or relationships to the expert review system of the secondary node platform. The expert review system is composed of experts in related fields who review and certify the new content. After expert certification, assign identification codes to new knowledge points and relationships and update them in the knowledge point relationship network. The update process includes adding new nodes and edges in the graph structure to ensure the integrity and accuracy of the relationship network.

[0107] In this embodiment, official textbooks can be effectively parsed to generate knowledge point entities with identification codes and establish a relationship network containing sequential, parallel, and comprehensive types. This relationship network provides a clear knowledge structure framework for teaching, which helps teachers better organize teaching content and students better master the association between knowledge points, improving learning effectiveness. At the same time, the incremental parsing and updating mechanism ensures that the relationship network can timely reflect the content of newly added teaching resources, maintaining its timeliness and accuracy.

[0108] In some embodiments, in step S102 above, based on the relationship network, the multi-platform question bank solving step is collected, authenticated by combining the identification code matching mechanism, synchronized to the basic knowledge space after authentication and marked as an extended knowledge point library, and the difficulty threshold of the knowledge point association combination is calculated by combining the mistake statistics, specifically including:

[0109] The multi-platform question bank solving step is split into atomic operation units and the knowledge point elements are extracted by calling the step semantic parser;

[0110] Based on the relationship network, the knowledge point elements are input into the identification matching engine for matching. If the matching fails, the expert review request process is triggered, the identification code is assigned to the new knowledge point after expert authentication, and the basic knowledge space is updated and marked as an extended knowledge point library;

[0111] Based on the authentication result, the atomic operation units are reorganized to generate a knowledge point association path, and the learner's mistake database is accessed to calculate the correctness of each step. When the correctness is lower than the preset correctness threshold, the corresponding knowledge point combination is automatically marked as a difficult point.

[0112] In this embodiment, the multi-platform question bank contains rich teaching resources, but the problem-solving steps are scattered and lack unified knowledge point identification, making it difficult to be directly used for teaching optimization. This embodiment aims to collect multi-platform question bank problem-solving steps based on the constructed knowledge point relationship network, authenticate through identification code matching mechanism, synchronize to the basic knowledge space and mark as an extension knowledge point library, and at the same time, combine the wrong question statistics to calculate the difficulty threshold of knowledge point association combination, to provide accurate difficulty analysis and targeted teaching suggestions for teaching.

[0113] Specifically, determine the multi-platform question bank that needs to collect problem-solving steps. These platforms can be common online education platforms, learning APPs, etc. Access the access rights and problem-solving step data interface of the question bank of each platform by technical docking. For example, by signing a data sharing agreement with the platform party, the text data or structured data of the problem-solving steps are obtained. Use data collection tools or scripts to regularly collect problem-solving steps from each platform question bank. The collection frequency can be set according to actual needs, such as daily, weekly, etc. The collected data should include question information, detailed description of problem-solving steps, problem-solving time, etc. Related information, and store the collected data in a temporary database for subsequent processing.

[0114] Develop or select a suitable step semantic parser, which should have the ability to semantically understand the problem-solving steps described in natural language. For example, the parser can identify semantic elements such as actions, objects, and conditions in the steps.

[0115] Input the collected problem-solving steps into the step semantic parser, which will split the problem-solving steps into atomic-level operation units according to semantic logic. For example, the problem-solving step of a math problem "first calculate the expression inside the parentheses, then perform multiplication" can be split into two atomic-level operation units "calculate the expression inside the parentheses" and "perform multiplication".

[0116] While splitting the atomic-level operation units, the parser extracts knowledge point elements from each operation unit. Knowledge point elements can be specific concepts, formulas, theorems, etc. For example, in the operation unit "calculate the expression inside the parentheses", the extracted knowledge point elements may be "four arithmetic operation rules" and the like.

[0117] Use the constructed knowledge point relationship network to build an identification matching engine. The engine can find matching knowledge point identification codes in the relationship network according to the characteristics of the knowledge point elements. Input the extracted knowledge point elements into the identification matching engine for matching. The engine finds the corresponding knowledge point identification code in the relationship network according to the semantic information and context relationship of the knowledge point elements. For example, if the extracted knowledge point element is "Pythagorean theorem", the engine finds the identification code corresponding to the theorem in the relationship network.

[0118] If the matching fails, it means that the knowledge point element can be a new knowledge point or a knowledge point not recorded in the existing relationship network. At this time, the expert review request process is triggered. The knowledge point element that fails to match and the related problem solving step information are sent to the expert review system. The expert review system is composed of teachers, education experts and other related disciplines, who will audit and evaluate the submitted content.

[0119] After expert certification, if it is confirmed that the knowledge point element is a new knowledge point, the expert will assign it a globally unique identification code. At the same time, the new knowledge point and its identification code are updated to the basic knowledge space and marked as an extension knowledge point library. The extension knowledge point library is used to store important knowledge points that are outside the teaching materials but appear in the actual problem solving process, providing a basis for the expansion of teaching content.

[0120] According to the identification matching result, the atomic operation unit is recombined according to the problem solving order to generate a knowledge point association path. For example, the problem solving process of a physics problem involves multiple knowledge points. Through identification matching, the knowledge point identification code corresponding to each operation unit is connected to form a complete knowledge point association path.

[0121] The generated knowledge point association path is connected to the learner's mistake database. The learner's mistake database records the error information of students in the problem solving process, including problem information, problem solving steps, error causes, etc.

[0122] For each step in the knowledge point association path, the correct rate of the learner at that step is calculated. The correct rate can be obtained by calculating the ratio of the number of students who correctly solved the problem to the total number of students who solved the problem. When the correct rate of a knowledge point combination corresponding to a step is lower than the preset correct rate threshold, the knowledge point combination is automatically marked as a difficult point. For example, if the preset correct rate threshold is 70%, if the problem solving step correct rate of a knowledge point combination in multiple problems is lower than 70%, the knowledge point combination is identified as a teaching difficulty.

[0123] In this embodiment, the problem solving steps of the multi-platform question bank can be effectively collected, and the knowledge point elements can be extracted and identified. New knowledge points are included in the basic knowledge space and marked as an extension knowledge point library. At the same time, the correct rate of the knowledge point association combination is calculated in combination with the learner's mistake database to accurately identify the teaching difficulty. This provides teachers with rich teaching resources and analysis basis, which helps teachers to conduct targeted teaching for difficult points and improve teaching quality; it also helps students to better understand their knowledge weaknesses and conduct targeted learning.

[0124] In some embodiments, in step S103, the learner's cross-platform problem-solving steps are captured in real time based on the relationship network, and an identification code matching mechanism is used for authentication. If authentication fails, the learner's problem-solving knowledge space with status markers is generated, which specifically includes:

[0125] The learner's cross-platform problem-solving step flow is captured in real time, and a step standardization processor is called to convert it into a structured operation sequence.

[0126] Based on the relationship network, the structured operation sequence is input into an identification matching engine for matching. If the matching fails, the expert system is triggered for rapid authentication, and when the authentication fails, the dependent prerequisite knowledge points are automatically located and marked as unmastered.

[0127] The matching results and status markers are fused, the knowledge point association path is constructed according to the problem-solving time sequence, and the traceable knowledge space unit with status markers is packaged.

[0128] In this embodiment, in the digital education environment, learners often practice problem-solving on multiple online learning platforms. However, these cross-platform problem-solving data are scattered and lack unified management, making it difficult to directly reflect the learner's knowledge mastery. This embodiment aims to capture the learner's cross-platform problem-solving steps in real time based on the constructed knowledge point relationship network, authenticate through the identification code matching mechanism, mark the unmastered prerequisite knowledge points, and generate the learner's problem-solving knowledge space with status markers, providing strong support for personalized teaching and learning diagnosis.

[0129] Specifically, a data docking mechanism is established with multiple mainstream online learning platforms. Through negotiation with platform providers, the problem-solving step data interface permission provided by the platform is obtained. For example, cooperation is reached with online mathematics practice platforms, physics simulation experiment platforms, etc., to ensure that the learner's problem-solving operation records on these platforms can be obtained in real time.

[0130] Based on data access, a special data capture tool is deployed. This tool can be a customized software based on web crawler technology, which can capture the learner's problem-solving step data from the interfaces of various platforms according to the preset rules and frequency. The captured data should include learner ID, problem information, problem-solving timestamp, and specific problem-solving operation description.

[0131] The captured problem-solving step flow data is transmitted to the intermediate data processing server in real time. The server uses an efficient data transmission protocol to ensure data integrity and timeliness. On the server, a dedicated data storage area is established for each learner, and the captured problem-solving step flow is stored in chronological order for subsequent processing.

[0132] A step standardization processor is developed, which has the ability to process problem-solving steps of different platforms and formats uniformly. A series of rules and templates are pre-set in the processor to identify and convert various elements in the problem-solving steps.

[0133] The real-time captured problem-solving step stream is input into the step standardization processor. The processor first performs text cleaning on the problem-solving steps to remove irrelevant characters, format information, etc. Then, according to the pre-set rules and templates, the problem-solving steps are split into a series of standardized operation units and arranged into a structured operation sequence according to the problem-solving order. For example, for the problem-solving step of a geometry problem "first draw a line segment AB, then draw a circle with A as the center and AB as the radius", the processor converts it into two structured operation units "draw line segment AB" and "draw a circle with A as the center and AB as the radius".

[0134] Using the constructed knowledge point relationship network, configure the identification matching engine. The engine stores the identification codes of all knowledge points and their relationship information, and can quickly find the corresponding knowledge point identification code according to the input operation unit. The structured operation sequence after standardization is input into the identification matching engine for matching. The engine analyzes each operation unit in turn, and according to the semantic information and context relationship of the operation unit, finds the matching knowledge point identification code in the knowledge point relationship network.

[0135] If a certain operation unit fails to match, it means that the knowledge point involved in the operation may not be explicitly identified in the existing relationship network, or the learner's operation has special circumstances. At this time, the expert system rapid authentication process is triggered. The operation unit that fails to match and related problem information, problem-solving context, etc. are sent to the expert system. The expert system is composed of teachers and education experts in related disciplines, who can quickly review and judge the submitted content.

[0136] After the expert system authentication, if it is confirmed that the knowledge point corresponding to the operation unit is not mastered by the learner, the system automatically locates the prerequisite knowledge points of the knowledge point. The positioning of the prerequisite knowledge points can be achieved by analyzing the logical relationship and dependency relationship in the knowledge point relationship network. For example, when learning the concept of derivative in higher mathematics, if the learner does not master the concept of limit, the system will mark the concept of limit as a prerequisite knowledge point not mastered.

[0137] According to the identification matching result and the state marking information, a knowledge point association path is constructed according to the problem solving sequence. For each structured operation unit, if the matching is successful, the corresponding knowledge point identification code is recorded; if the matching fails and the pre-knowledge point is marked as a state of not being mastered, the identification code and state information of the pre-knowledge point are also recorded. The knowledge point information corresponding to all operation units is connected according to the problem solving order to form a complete knowledge point association path. For example, in the problem solving process of a chemical experiment question, multiple knowledge points are involved in the operation. Through the construction of the knowledge point association path, the knowledge points applied by the learner in the problem solving process in sequence and the possible knowledge gaps can be clearly seen.

[0138] The constructed knowledge point association path is encapsulated as a traceable knowledge space unit with a state mark. The knowledge space unit not only contains the knowledge point association path, but also records the problem solving time of the learner, the source of the question, the detailed information of the problem solving steps, etc., so as to be traced and analyzed subsequently. For example, the teacher can understand the learning process and problems of the learner at a certain knowledge point by checking the knowledge space unit, and provide a basis for individualized tutoring. At the same time, the encapsulated knowledge space unit is stored in the personal knowledge space database of the learner, which is convenient for calling and updating at any time.

[0139] In this embodiment, the problem solving steps of the learner across platforms can be captured in real time, and standardized processing and identification matching can be performed. For unauthenticated operations, the pre-knowledge point not mastered state can be accurately marked, the knowledge point association path with state marking can be constructed, and the traceable knowledge space unit can be encapsulated. This provides the teacher with a comprehensive and detailed analysis of the knowledge mastery of the learner, helps the teacher to implement individualized teaching strategies and improve the teaching quality; it also helps the learner to better understand his own learning situation and to make targeted knowledge supplement and learning improvement.

[0140] In some embodiments, in the step S104, the mapping of the learner and the standard answer path by the path comparator for solving the question, the calculation of the feature state value based on the multi-dimensional analysis model and the statistics of the knowledge point mastery rate, specifically include:

[0141] For solving the question, the learner's problem solving path and the standard answer path are aligned by the path comparator according to the logic segment, and the sliding window algorithm is used to calculate the overlap ratio of the segment knowledge point identification code;

[0142] The overlap ratio is determined to be up to standard and the A dimension state is marked, the path endpoint identification code consistency is detected and the B dimension state is marked, the final answer correctness is verified and the C dimension state is marked, and the extension knowledge point library identification code is scanned and the D dimension state is marked;

[0143] Based on the A-dimension state, the B-dimension state, the C-dimension state and the D-dimension state, the historical question data is aggregated to construct a knowledge point-application matrix.

[0144] By traversing the knowledge point-application matrix, the frequency of each knowledge point being marked as mastered and the total frequency of application are counted to calculate the knowledge point mastery rate.

[0145] In this embodiment, in the process of education and teaching, the answer question is an important question type for testing the comprehensive application ability of learners' knowledge. However, the traditional teaching evaluation method often cannot comprehensively and accurately analyze the knowledge point mastery of learners in the answer question. This embodiment aims at the answer question, maps the learner's problem-solving path with the standard answer path through the path comparator, calculates the characteristic state value by using the multi-dimension analysis model, and counts the knowledge point mastery rate, so as to provide more accurate teaching feedback for teachers and help learners better understand their weak points.

[0146] Specifically, in an online learning platform or examination system, the operation steps of learners in answering questions are recorded in real time. When a learner starts to answer a question, the system records the operation content of each step in time sequence, such as the input text, the drawn graphics, the selected options, etc. These operation steps are arranged according to the logical relationship to form the problem-solving path of the learner. For example, in a mathematical geometry proof question, the problem-solving path of the learner may include steps such as "drawing auxiliary lines", "quoting theorems", "conducting reasoning", etc.

[0147] The standard answer path is pre-prepared by subject experts according to the requirements of the question and the knowledge point system. The standard answer path describes the standard steps of answering the question in detail and the knowledge points involved. Taking the mathematical geometry proof question as an example again, the standard answer path will clearly indicate the theorems, formulas and reasoning processes that need to be used at each step.

[0148] A path comparator is developed or selected, which has the ability to logically segment and align the two paths. The logical segments can be segmented according to the semantics and functions of the problem-solving steps, for example, in a mathematical question, the proof process can be divided into logical segments such as "introducing conditions", "conducting reasoning", "drawing conclusions", etc.

[0149] The learner's problem-solving path and the standard answer path are input into the path comparator, which divides the two paths into corresponding logical segments according to the pre-set logical segment division rules. Then, the sliding window algorithm is used to compare the knowledge point identification codes in each logical segment. The sliding window algorithm slides on the logical segments of the two paths, and calculates the overlap ratio of the knowledge point identification codes in the window. For example, in a certain logical segment, the knowledge point identification codes involved in the standard answer path are A, B, C, and the knowledge point identification codes involved in the learner's problem-solving path are A, B, D, then the overlap ratio of the knowledge point identification codes in this logical segment is 2 / 3.

[0150] According to the characteristics of the subject and the difficulty of the question, the passing standard of the overlap ratio of knowledge point identification code is set in advance. For example, for simple answer questions, the passing standard can be set to an overlap ratio of not less than 80%; for complex answer questions, the passing standard can be set to not less than 60%. Compare the calculated overlap ratio of knowledge point identification code in each logical segment with the passing standard. If the overlap ratio reaches or exceeds the passing standard, mark the A dimension state of the logical segment as "passing"; otherwise, mark it as "not passing".

[0151] Extract the knowledge point identification code of the path endpoint from the learner's problem-solving path and the standard answer path. The knowledge point identification code of the path endpoint usually represents the core knowledge point finally applied in the answer question. Compare whether the learner's problem-solving path endpoint identification code and the standard answer path endpoint identification code are consistent. If consistent, mark the B dimension state as "consistent"; otherwise, mark it as "inconsistent".

[0152] According to the requirements of the question and the standard answer, judge the learner's final answer. Artificial judgment or automatic judgment system (such as for questions with clear answers such as multiple choice questions and fill-in-the-blank questions) can be used for judgment. If the learner's final answer is correct, mark the C dimension state as "correct"; otherwise, mark it as "incorrect".

[0153] Establish an extended knowledge point library, which stores the knowledge point identification codes that may be involved in the learner's problem-solving process but are not explicitly embodied in the standard answer path. These knowledge points may be a supplement or expansion of the standard answer, which can help to more comprehensively evaluate the learner's knowledge mastery. Scan the knowledge point identification codes in the learner's problem-solving path and check if there are matching identification codes in the extended knowledge point library. If there are, mark the D dimension state as "involving extended knowledge points"; otherwise, mark it as "not involving extended knowledge points".

[0154] Collect the learner's problem-solving path data and corresponding multi-dimensional state marks (A, B, C, D dimension states) in different times and different questions. These data can be stored in a database for subsequent analysis and processing. Aggregate historical question data according to knowledge point identification codes. Integrate all problem-solving path data and multi-dimensional state marks involving the same knowledge point to form a data set with knowledge points as the dimension.

[0155] A knowledge point-application matrix is constructed, with rows representing different knowledge points and columns representing different application scenarios (i.e., different questions). The elements in the matrix can record the multi-dimensional state marker information of the knowledge point in the question. According to the aggregated historical question data, the multi-dimensional state marker information of each knowledge point in different questions is filled into the knowledge point-application matrix. For example, for the solving of knowledge point X in question Y, the corresponding A, B, C, and D dimensional state markers are filled into the corresponding position of the matrix.

[0156] The knowledge point-application matrix is traversed to count the frequency of each knowledge point being marked as "mastered". In this embodiment, the judgment standard of "mastered" can be set according to actual conditions. For example, when the A, B, and C dimensional states are "up to standard", "consistent", and "correct", it is considered that the knowledge point is mastered in the question. The number of times each knowledge point is marked as "mastered" in all questions is counted. At the same time, the total frequency of application of each knowledge point in all questions, i.e., the total number of times the knowledge point appears in the knowledge point-application matrix, is counted.

[0157] For each knowledge point, the frequency of being marked as "mastered" is divided by the total frequency of application to obtain the mastery rate of the knowledge point. For example, knowledge point X is applied in 10 questions, of which 6 are marked as "mastered", so the mastery rate of knowledge point X is 6 / 10 = 60%.

[0158] In this embodiment, the solving path of the learner can be comprehensively and accurately analyzed for answer questions, and the mastery rate of each knowledge point can be calculated through multi-dimensional state marking and knowledge point-application matrix construction. This provides detailed and intuitive teaching feedback for teachers, helping teachers understand the mastery of learners on different knowledge points, so as to provide targeted teaching guidance and course design. At the same time, learners can also understand their weak points according to the knowledge point mastery rate, and carry out targeted learning and review to improve learning effect.

[0159] In some embodiments, in step S105, for a selected true-false question, a same-type answer question is matched through a question type correlator, and a mistake type is inferred according to the characteristic state of the associated answer question, specifically including:

[0160] The tail identification code of the solving path of the true-false question is extracted through the question type correlator, and all answer questions containing the same tail identification code are matched in the question type fingerprint library using the tail identification code as the key;

[0161] The latest characteristic state record of the learner on the matched answer question is obtained, the proportion of each characteristic state is counted, and the dominant characteristic state is determined;

[0162] Compare the results of the selection judgment questions with the dominant feature states, and when there is a contradiction, according to the preset rules to deduce the error type, and generate targeted learning remediation instructions.

[0163] In this embodiment, in the education evaluation system, selection judgment questions and answer questions are common types of questions. Selection judgment questions are usually used to quickly examine the memory and understanding of learners on basic knowledge, while answer questions focus more on the comprehensive application ability and thinking process of learners. However, only through the results of selection judgment questions, it is often difficult to deeply understand the specific reasons for the errors of learners. This embodiment aims to selection judgment questions, match the same type of answer questions through the question type correlator, and then deduce the error type according to the feature state of the associated answer questions, so as to provide more accurate learning remediation suggestions for learners and improve learning effectiveness.

[0164] Specifically, a question type correlator is developed, which has the functions of extracting the tail identification code of the solution path of the question and matching in the question type fingerprint library. The question type fingerprint library is a pre-established data storage system, which stores a large amount of solution path information of answer questions and corresponding question feature identifiers, including tail identification code of solution path, etc. The meaning of the tail identification code of the solution path is clear. The tail identification code can represent the identification of the key knowledge points or solution steps involved in the last solution path. For example, in a selection judgment question about historical events, the tail identification code may correspond to the core knowledge points of the historical event, such as the time, place, main characters, etc.

[0165] When the learner completes a selection judgment question, the question type correlator extracts the tail identification code of the solution path of the question. The solution path can be generated by recording the operation steps and thinking process of the learner during the answering process, such as recording the options selected by the learner, the analysis of the question, etc. Take the extracted tail identification code as the key to search and match in the question type fingerprint library. Find all the answer questions containing the same tail identification code. These answer questions have certain relevance with the selection judgment question in terms of key knowledge points or solution steps involved. For example, if the tail identification code of the selection judgment question corresponds to a certain mathematical theorem, then the matched answer questions may all be application questions around the theorem.

[0166] After the learner completes the matched answer questions, the system records the learner's answering situation on these questions, including answering time, answering steps, final answer, etc. These information constitutes the feature state record of the learner on the answer questions. The feature state record of the learner is sorted and stored in the database. The database can be classified and stored according to the fields of learner ID, question ID, etc.

[0167] According to the answers, the characteristic states are classified. For example, the characteristic states can be classified into "clear problem-solving approach", "incorrect problem-solving steps", "unfamiliar knowledge application", etc.

[0168] For each learner's characteristic state record on matching answers, the number of occurrences of each type of characteristic state is counted, and its proportion in the total number of records is calculated. For example, if a learner has 6 questions showing the "incorrect problem-solving steps" characteristic state in 10 matching answer questions, the proportion of this characteristic state is 60%.

[0169] A threshold value for the proportion of a characteristic state is set in advance, for example 50%. Compare the proportion of each type of characteristic state with the threshold value. If the proportion of a certain characteristic state exceeds the threshold value, then the characteristic state is determined as the dominant characteristic state. For example, if the proportion of "incorrect problem-solving steps" is 60%, which exceeds the threshold value of 50%, then "incorrect problem-solving steps" is the dominant characteristic state of the learner.

[0170] After the learner completes the selection judgment questions, the system records the results of his answers, i.e. whether the selected option is correct.

[0171] According to the characteristics of the selection judgment questions and the knowledge points investigated, establish comparison rules between the results of the selection judgment questions and the dominant characteristic states. For example, if the selection judgment question investigates the memory of a certain knowledge point, and the dominant characteristic state is "unfamiliar knowledge application", then when the learner makes a mistake on the selection judgment question, there may be a problem of inaccurate memory or insufficient understanding.

[0172] Compare the learner's selection judgment question results with the dominant characteristic state. Determine whether there is a contradiction between the two. For example, if the correct answer to the selection judgment question requires the use of a certain knowledge point for reasoning, and the learner selects the wrong answer to the selection judgment question, and the dominant characteristic state is "unclear problem-solving approach", then it can be considered that there is a contradiction between the two.

[0173] According to common answer mistakes, develop preset rules. For example, when the selection judgment question result is wrong and the dominant characteristic state is "incorrect problem-solving steps", the possible reverse deduction of the mistake type is "deviation in understanding and mastery of problem-solving steps"; when the selection judgment question result is wrong and the dominant characteristic state is "unfamiliar knowledge application", the possible reverse deduction of the mistake type is "insufficient application ability of knowledge points".

[0174] According to the comparison results and preset rules, the learner's mistake type on the selection judgment question is deduced.

[0175] A remedial strategy library is established, which stores learning remedial strategies for different types of errors. For example, for the error type of "deviation in understanding and mastering problem-solving steps", the remedial strategies can include providing detailed explanation of problem-solving steps, conducting similar problem-solving step exercises, etc.; for the error type of "insufficient application ability of knowledge points", the remedial strategies can include providing practical application cases of knowledge points, organizing group discussions on application of knowledge points, etc.

[0176] According to the error type deduced, the corresponding remedial strategy is selected from the remedial strategy library to generate targeted learning remedial instructions. For example, if the error type is "deviation in understanding and mastering problem-solving steps", the generated learning remedial instructions can be "watch the problem-solving step explanation video of the corresponding knowledge point of the selected judgment question, and complete 5 similar problem-solving step exercises".

[0177] The learning remedial instructions are delivered to the learners in a suitable way. For example, the instructions can be pushed to the learners through the message system of the online learning platform, email, etc.

[0178] After receiving the learning remedial instructions, the learners follow the instructions to learn and practice, so as to make up for their knowledge gaps and insufficient problem-solving ability in the selected judgment questions.

[0179] In this embodiment, for the selected judgment questions, the same type of answer questions are matched through the question type associator, the error type is deduced according to the characteristic state of the learners in the answer questions, and targeted learning remedial instructions are generated. This helps teachers and learners better understand the reasons for the errors of the learners in the selected judgment questions, provides personalized learning support for the learners, and improves the learning effect and performance of the learners. At the same time, it also provides a more scientific and accurate basis for teaching evaluation and teaching improvement.

[0180] In some embodiments, in the step S106, the high coincidence degree question types are clustered based on the head and tail information of the problem-solving path, the class data is projected to the stage knowledge space, the multi-dimensional heat map is rendered to visualize the knowledge defect distribution according to the knowledge point mastery rate, and the weight relationship between the knowledge point mastery degree and the problem-solving characteristics is mined through the Apriori algorithm to generate knowledge point mastery path optimization suggestions, which specifically include:

[0181] The question head and tail knowledge point identification codes are extracted by the path characteristic extractor, and the density clustering algorithm is used to classify the questions with a coincidence degree exceeding a preset coincidence degree threshold into the same question type group, and a unique clustering identification code is generated for each question type group;

[0182] Based on the unique cluster identification code, a set of periodic knowledge points is extracted according to the teaching progress, a class student-knowledge point mastery rate matrix is constructed, and the matrix data of the class student-knowledge point mastery rate matrix is mapped to a three-dimensional knowledge space coordinate system;

[0183] Based on the three-dimensional knowledge space coordinate system, a mastery rate classification color mapping rule is set, and a transparency parameter is used to represent the number density to generate an interactive heat map, which is used to display the knowledge defect distribution;

[0184] Based on the interactive heat map, the student problem solving characteristics and the knowledge point mastery rate are converted into a transaction item set, the strong association rules are mined by improving the Apriori algorithm, and the knowledge point mastery path optimization suggestions are output.

[0185] In this embodiment, in the education and teaching, accurately grasping the mastery of different knowledge points by the students in the class and optimizing the knowledge point mastery path are crucial for improving the teaching quality and the learning effect of the students. The traditional knowledge point mastery evaluation method often lacks in-depth analysis of the problem type and problem solving characteristics, and it is difficult to intuitively display the knowledge defect distribution and mine the internal relationship between the knowledge point mastery degree and the problem solving characteristics. This embodiment aims to cluster the high coincidence degree question types through the head and tail information of the problem solving path, combine the class data visualization and data mining technology, and provide more accurate knowledge point mastery path optimization suggestions for teachers and students.

[0186] A development path feature extractor is developed, which can accurately extract the head and tail knowledge point identification codes from the problem solving path of the question. The problem solving path can be generated by recording the operation steps, thinking logic and other information during the student's answering process. For example, in a mathematical application question, the problem solving path may include steps such as analysis of the question conditions, construction of the problem solving approach, and calculation of the final answer, and the head and tail knowledge point identification codes correspond to the key knowledge points involved at the beginning and end of the problem solving.

[0187] The coding rules of the knowledge point identification code are set to ensure that each knowledge point has a unique and identifiable identification code. For example, the identification code can be generated by combining subject classification, chapter number, knowledge point serial number, etc.

[0188] The density clustering algorithm (such as DBSCAN algorithm) is selected to cluster the questions. The density clustering algorithm can discover clusters of arbitrary shape and has certain robustness to noise data, which is suitable for processing the complex distribution of question data.

[0189] According to the teaching needs and the characteristics of the questions, the coincidence threshold is set in advance. The coincidence degree can be measured by comparing the similarity of the head and tail knowledge point identification codes of the questions. For example, when the head and tail knowledge point identification codes of two questions have more than a certain proportion (such as 70%) of coincidence, it is considered that the two questions have high coincidence degree.

[0190] All questions are input into the path feature extractor to extract the head and tail knowledge point identification code. Then, the density clustering algorithm is used to cluster the questions, and the questions with a coincidence degree exceeding a preset threshold are classified into the same question type group.

[0191] A unique cluster identification code is generated for each question type group. The identification code can include cluster number, cluster timestamp, etc. generated by the clustering algorithm to ensure the uniqueness and traceability of the identification code. The generated cluster identification code is stored in association with the corresponding question type group information.

[0192] According to the teaching plan and actual teaching progress, determine the knowledge point set involved in the current stage. For example, in the teaching process of a semester, different teaching stages are divided according to the teaching week or chapter order, and the knowledge points that need to be mastered in each stage are clearly defined. From all the knowledge points, the knowledge points that need to be examined in the current stage are selected to form a set of stage knowledge points.

[0193] Collect the class students' answer data on each knowledge point, including the correct rate, answer time, problem solving steps, etc. According to these data, calculate the mastery rate of each student on each knowledge point, for example, the mastery rate can be measured by the average score rate of the student on the knowledge point related questions. The class students are taken as rows, and the set of stage knowledge points is taken as columns to construct a class student-knowledge point mastery rate matrix. Each element in the matrix represents the mastery rate of the corresponding student on the corresponding knowledge point.

[0194] A three-dimensional knowledge space coordinate system is constructed, in which three dimensions represent different dimension information. For example, one dimension can be set as the knowledge point difficulty level, one dimension can be set as the chapter to which the knowledge point belongs, and the other dimension can be set as the knowledge point mastery rate.

[0195] Map the data in the class student-knowledge point mastery rate matrix to the three-dimensional knowledge space coordinate system. The knowledge point mastery rate data of each student forms a data point in the coordinate system, and through the position and attribute information of the data point, the mastery of the class students on different knowledge points can be intuitively displayed.

[0196] According to the teaching needs and actual situation, determine the grading standard of knowledge point mastery rate. For example, the mastery rate can be divided into excellent (80%-100%), good (60%-79%), pass (40%-59%), and fail (0%-39%) four levels.

[0197] Different colors are assigned to each mastery rate level to form a color mapping rule. For example, the excellent level corresponds to green, the good level corresponds to blue, the pass level corresponds to yellow, and the fail level corresponds to red. Through color changes, the distribution of different knowledge point mastery rates can be intuitively displayed.

[0198] In the three-dimensional knowledge space coordinate system, the number of data points in each region, i.e., the corresponding number of students, is counted to calculate the population density. The transparency parameter of the data points is set according to the population density. The higher the population density of a region, the lower the transparency of the data points, and the brighter the color. The lower the population density of a region, the higher the transparency of the data points, and the lighter the color. Through the change of transparency, the concentration degree of the students in the class on different knowledge points can be intuitively displayed.

[0199] Select appropriate visualization tools (such as D3.js, Tableau, etc.) to generate interactive heat maps. Interactive heat maps allow users to zoom, rotate, pan, and other operations through mouse operations to observe the data in more detail. The data points with set color mapping rules and transparency parameters are displayed in the visualization tool to generate an interactive heat map. The heat map can intuitively display the mastery of students in the class on different knowledge points and the distribution of knowledge defects.

[0200] Extract problem-solving features from student answer data, such as problem-solving time, number of problem-solving steps, whether to use auxiliary tools, etc. Convert student problem-solving features and knowledge point mastery rates into transaction item sets. Each student's answer record can be regarded as a transaction, which contains the student's problem-solving features and the mastery rate information of the corresponding knowledge point in the transaction.

[0201] Improve the traditional Apriori algorithm to improve the efficiency and accuracy of mining. For example, you can use a hash-based technique to reduce the number of candidate sets generated, or use parallel computing methods to speed up the algorithm's operation.

[0202] Use the improved Apriori algorithm to mine transaction item sets and find strong association rules. Strong association rules indicate that there is a strong correlation between knowledge point mastery and problem-solving features. For example, it may be found that students who solve problems quickly and clearly have a higher mastery rate on a certain knowledge point.

[0203] Based on the strong association rules mined, combined with teaching goals and the actual situation of students, generate knowledge point mastery path optimization suggestions. For example, if it is found that students have a low mastery rate on a certain knowledge point, and it is related to unclear problem-solving steps, the optimization suggestions can include strengthening the explanation and practice of problem-solving steps for that knowledge point.

[0204] Push the generated knowledge point mastery path optimization suggestions to teachers and students to allow teachers to adjust teaching strategies and students to improve learning methods.

[0205] In this embodiment, high coincidence types can be clustered based on the head and tail information of the problem solving path, class data can be projected to the stage knowledge space, and the knowledge defect distribution can be intuitively displayed through an interactive heat map. Meanwhile, the weight relationship between the knowledge point mastery degree and the problem solving feature is mined by using the improved Apriori algorithm, and the targeted knowledge point mastery path optimization suggestion is generated. This helps teachers more accurately understand the knowledge mastery of students in the class, adjust the teaching strategy, and improve the teaching quality; it also helps students find their weak points in knowledge, optimize their learning methods, and improve their learning effect.

[0206] Referring to Figure 2 An embodiment of the present application provides a teaching optimization system 2 based on a knowledge space, and the system 2 specifically comprises:

[0207] A first teaching optimization module 201 is configured to parse a secondary node platform based on a national industrial internet identifier, parse official teaching materials to generate knowledge point entities with an identifier code, and establish a relationship network including a sequential type, a parallel type and a comprehensive type.

[0208] A second teaching optimization module 202 is configured to collect problem solving steps of a multi-platform question bank based on the relationship network, authenticate by combining an identifier code matching mechanism, synchronize to a basic knowledge space after authentication and mark as an extended knowledge point library, and calculate a difficulty threshold of a knowledge point association combination by combining mistake statistics.

[0209] A third teaching optimization module 203 is configured to capture learning steps across platforms in real time based on the relationship network, authenticate by combining an identifier code matching mechanism, mark as a pre-knowledge point unmastered state if authentication fails, and generate a learner problem solving knowledge space with a state mark.

[0210] A fourth teaching optimization module 204 is configured to map a learner and a standard answer path by a path comparator for a solved question, calculate a feature state value based on a multi-dimensional analysis model and calculate a knowledge point mastery rate.

[0211] A fifth teaching optimization module 205 is configured to match a same-type solved question by a question type associator for a selection and judgment question, and then infer a mistake type according to a feature state of the associated solved question.

[0212] A sixth teaching optimization module 206 is configured to cluster high coincidence types based on head and tail information of a problem solving path, project class data to a stage knowledge space, render a multi-dimensional heat map to visualize knowledge defect distribution according to a knowledge point mastery rate, and generate a knowledge point mastery path optimization suggestion by mining a weight relationship between a knowledge point mastery degree and a problem solving feature by using an Apriori algorithm.

[0213] It can be understood that, as Figure 1The content of the illustrated knowledge space-based teaching optimization method embodiment is applicable to the present knowledge space-based teaching optimization system embodiment, the function implemented by the present knowledge space-based teaching optimization system embodiment is the same as that of the knowledge space-based teaching optimization method embodiment as Figure 1 The present knowledge space-based teaching optimization system embodiment achieves the same beneficial effects as the knowledge space-based teaching optimization method embodiment as Figure 1 The present knowledge space-based teaching optimization system embodiment achieves the same beneficial effects as the knowledge space-based teaching optimization method embodiment as

[0214] It should be noted that the information interaction and execution process between the above systems are based on the same concept as the method embodiments, and the specific functions and technical effects brought by them can be referred to the method embodiments part for details, which will not be described here.

[0215] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above functional units and modules is exemplified, and in actual application, the above functions can be completed by different functional units and modules according to needs, that is, the internal structure of the system is divided into different functional units or modules to complete all or part of the above described functions. The functional units and modules in the embodiments can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or software. In addition, the specific names of the functional units and modules are only for easy distinction, and do not limit the protection scope of the present application. The specific working process of the units and modules in the system can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0216] With reference to Figure 3 The present embodiment further provides a computer device 3, comprising a memory 302 and a processor 301 and a computer program 303 stored in the memory 302, when the computer program 303 is executed on the processor 301, the knowledge space-based teaching optimization method is realized as any one of the above methods.

[0217] The computer device 3 can be a desktop computer, a notebook computer, a palm computer, a cloud server and other computing devices. The computer device 3 can include, but is not limited to, a processor 301, a memory 302. Those skilled in the art can understand that Figure 3 The computer device 3 is only an example and does not limit the computer device 3, which can include more or fewer components than shown, or combine certain components, or different components, for example, it can also include input / output devices, network access devices, etc.

[0218] The processor 301 can be a central processing unit (CPU), and can also be other general-purpose processors, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0219] The memory 302 can be an internal storage unit of the computer device 3 in some embodiments, for example, a hard disk or a memory of the computer device 3. The memory 302 can also be an external storage device of the computer device 3 in other embodiments, for example, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the memory 302 can include both the internal storage unit and the external storage device of the computer device 3. The memory 302 is used to store an operating system, an application program, a boot loader, data, and other programs, for example, program codes of the computer program, etc. The memory 302 can also be used to temporarily store data that has been output or is to be output.

[0220] The embodiment of the present application further provides a computer readable storage medium, which stores a computer program, and the computer program is run by a processor to implement the knowledge space-based teaching optimization method according to any one of the above methods.

[0221] In this embodiment, the integrated unit, if implemented in the form of a software functional unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the computer program for instructing the relevant hardware to complete all or part of the processes in the above-described embodiment methods can be stored in a computer readable storage medium. The computer program can be executed by a processor to implement the steps of the above-mentioned various method embodiments. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer readable medium at least includes any entity or device capable of carrying the computer program code to the photographing device / terminal equipment, recording medium, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium. For example, U disk, mobile hard disk, magnetic disk or optical disk, etc. In some jurisdictions, according to legislation and patent practice, the computer readable medium can not be an electrical carrier signal and a telecommunication signal.

[0222] In the above embodiments, the description of each embodiment has its own focus, and the parts not described or recorded in detail in a certain embodiment can be referred to the relevant description of other embodiments.

[0223] Those of ordinary skill in the art can understand that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are implemented in hardware or software depends on the specific application and design constraints of the technical solutions. 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 the present application.

[0224] In the embodiments disclosed in the present application, it should be understood that the disclosed apparatus / terminal equipment and methods can be implemented in other ways. For example, the apparatus / terminal equipment embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division, and actual implementation can have another division manner. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interface, device or unit, and can be electrical, mechanical or other forms.

[0225] The units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may also be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

Claims

1. A teaching optimization method based on knowledge space, characterized in that: The method specifically includes: Based on the National Industrial Internet Identifier Resolution Secondary Node Platform, official textbooks are parsed to generate knowledge point entities with identification codes, and a relationship network including sequential, parallel, and comprehensive types is established; Based on the relationship network, we collect the problem-solving steps of question banks from multiple platforms and authenticate them with the identification code matching mechanism. After authentication, we synchronize them to the basic knowledge space and mark them as extended knowledge point libraries. At the same time, we calculate the difficulty threshold of the knowledge point association combination based on the statistics of wrong questions. Based on the relationship network, learners' cross-platform problem-solving steps are captured in real time and authenticated by combining the identification code matching mechanism. If the authentication fails, the learners are marked as not having mastered the prerequisite knowledge points, and a problem-solving knowledge space with status marks is generated for the learners. For answering questions, the path matcher maps the learner's path to the standard answer, calculates the feature state value based on the multi-dimensional analysis model, and calculates the knowledge point mastery rate; For multiple-choice questions, the question type associater is used to match the answer questions of the same question type, and then the error type is inferred based on the characteristic status of the associated answer questions; Based on the information at the beginning and end of the problem-solving path, we cluster the question types with high overlap, project the class data into the staged knowledge space, render a multi-dimensional heat map to visualize the distribution of knowledge deficiencies according to the mastery rate of knowledge points, and use the Apriori algorithm to explore the weighted relationship between the mastery degree of knowledge points and the problem-solving characteristics, and generate optimization suggestions for the knowledge point mastery path.

2. The method according to claim 1, characterized in that Based on the National Industrial Internet Identifier Resolution Secondary Node Platform, the official textbooks are parsed to generate knowledge point entities with identification codes, and a relationship network including sequential, parallel and comprehensive types is established, specifically including: Through the semantic segmentation module of the National Industrial Internet Identifier Resolution Secondary Node Platform, official textbooks are parsed to generate independent knowledge units, and each knowledge unit is assigned a globally unique identification code; Based on the logical structure of the textbook, a causal chain analyzer is used to identify sequential relationships, while a semantic similarity matcher is used to extract parallel relationships. The cross-chapter association miner is then used to generate comprehensive relationships to form a relationship network. Create entity-relationship-entity triple identification codes for sequential and parallel relationships, and generate multi-entity and multi-relationship combination identification codes for comprehensive relationships; When new teaching resources are added, the incremental parser is triggered to extract new knowledge points or relationships, which are submitted to the expert review system of the secondary node platform. After certification, an identification code is assigned and the relationship network is updated.

3. The method according to claim 1, characterized in that Based on the relationship network, the problem-solving steps of the multi-platform question bank are collected, and the identification code matching mechanism is used for authentication. After authentication, the steps are synchronized to the basic knowledge space and marked as the extended knowledge point library. At the same time, the difficulty threshold of the knowledge point association combination is calculated based on the wrong question statistics. Specifically, it includes: Collect problem-solving steps from multiple platforms, call the step semantic parser to split them into atomic operation units and extract knowledge point elements; Based on the relationship network, the knowledge point elements are input into the identification matching engine for matching. If the match fails, the expert review request process is triggered. After expert certification, the newly added knowledge point is assigned an identification code, which is synchronously updated to the basic knowledge space and marked as an extended knowledge point library. Based on the certification results, the atomic operation units are reorganized to generate knowledge point association paths, and the learner's wrong question database is connected to count the accuracy of each step. When the accuracy is lower than the preset accuracy threshold, the corresponding knowledge point combination is automatically marked as difficult.

4. The method according to claim 1, wherein Based on the relationship network, the learner's cross-platform problem-solving steps are captured in real time, and authentication is performed in combination with the identification code matching mechanism. If the authentication fails, the learner is marked as not having mastered the previous knowledge points, and a learner's problem-solving knowledge space with status marks is generated. Specifically, it includes: Capture learners' cross-platform problem-solving steps in real time and call a step standardization processor to convert them into a structured operation sequence; Based on the relationship network, the structured operation sequence is input into the identification matching engine for matching. If the match fails, the expert system will trigger rapid authentication. If the authentication fails, the dependent pre-knowledge points will be automatically located and marked as not mastered. The matching results and status tags are integrated to build the knowledge point association path according to the problem-solving sequence, and encapsulate it into a traceable knowledge space unit with status tags.

5. The method according to claim 1, wherein For the answer questions, the path matcher maps the learner's path to the standard answer, calculates the feature state value based on the multi-dimensional analysis model, and calculates the knowledge point mastery rate, specifically including: For problem-solving questions, the path comparer aligns the learner's problem-solving path with the standard answer path in logical segments, and uses a sliding window algorithm to calculate the overlap ratio of knowledge point identification codes within the segment; Determine whether the overlap ratio meets the standard and mark the A-dimension status; check the consistency of the path end identification code and mark the B-dimension status; verify the correctness of the final answer and mark the C-dimension status; scan the extended knowledge point library identification code and mark the D-dimension status; Aggregate historical question data based on the A-dimension status, B-dimension status, C-dimension status, and D-dimension status to build a knowledge point-application matrix; By traversing the knowledge point-application matrix, the frequency of each knowledge point being marked as mastered and the total frequency of application are counted to calculate the knowledge point mastery rate.

6. The method according to claim 1, characterized in that For the selection and judgment questions, the question type associator is used to match the answer questions of the same question type, and then the error type is inferred based on the characteristic state of the associated answer questions. Specifically, the following steps are performed: Extract the tail identification code of the solution path of the judgment question through the question type associator, and use the tail identification code as the key to match all the answer questions with the same tail identification code in the question type fingerprint library; Obtain the learner's latest feature state records on the matching answer questions, calculate the proportion of each feature state and determine the dominant feature state; Compare the results of the judgment questions with the dominant feature status. When the two are inconsistent, infer the error type according to the preset rules and generate targeted learning remedial instructions.

7. The method according to any one of claims 1 to 6, characterized in that The method clusters highly overlapping question types based on the head and tail information of the problem-solving path, projects class data into a staged knowledge space, renders a multi-dimensional heat map to visualize the distribution of knowledge deficiencies based on the knowledge point mastery rate, and uses the Apriori algorithm to mine the weighted relationship between the knowledge point mastery level and the problem-solving characteristics to generate optimization suggestions for the knowledge point mastery path. Specifically, it includes: The path feature extractor extracts the knowledge point identification codes at the beginning and end of the questions, and uses a density clustering algorithm to classify questions with a degree of overlap exceeding a preset threshold into groups of the same question type, generating a unique cluster identification code for each question type group. Based on the unique cluster identification code, a set of staged knowledge points is extracted according to the teaching progress, and a class student-knowledge point mastery rate matrix is ​​constructed. The matrix data of the class student-knowledge point mastery rate matrix is ​​mapped to a three-dimensional knowledge space coordinate system; Based on the three-dimensional knowledge space coordinate system, color mapping rules for mastery rate classification are set, and transparency parameters are used to represent the population density to generate an interactive heat map, which is used to display the distribution of knowledge deficiencies. Based on the interactive heat map, the students' problem-solving characteristics and knowledge point mastery rates are converted into transaction item sets. The strong association rules are mined through the improved Apriori algorithm, and optimization suggestions for the knowledge point mastery path are output.

8. A teaching optimization system based on knowledge space, characterized by: The system specifically includes: The first teaching optimization module is used to parse official textbooks to generate knowledge point entities with identification codes based on the National Industrial Internet Identifier Resolution Secondary Node Platform, and establish a relationship network that includes sequential, parallel, and comprehensive types; The second teaching optimization module is used to collect problem-solving steps from question banks on multiple platforms based on the relationship network, authenticate them through the identification code matching mechanism, synchronize them to the basic knowledge space after authentication and mark them as extended knowledge point libraries. At the same time, the difficulty threshold of the knowledge point association combination is calculated based on the statistics of wrong questions. The third teaching optimization module is used to capture learners' cross-platform problem-solving steps in real time based on the relationship network, and authenticate them through the identification code matching mechanism. If they fail the authentication, they are marked as not having mastered the prerequisite knowledge points, and a problem-solving knowledge space with status tags is generated for the learners. The fourth teaching optimization module is used to map learners' paths to standard answers through a path matcher for solving questions, calculate feature state values ​​based on a multi-dimensional analysis model, and calculate knowledge point mastery rates; The fifth teaching optimization module is used to match the same type of answer questions with the question type associater for multiple-choice questions, and then infer the error type based on the characteristic status of the associated answer questions; The sixth teaching optimization module is used to cluster highly overlapping question types based on the head and tail information of the problem-solving path, project class data into a staged knowledge space, render a multi-dimensional heat map to visualize the distribution of knowledge deficiencies according to the mastery rate of knowledge points, and use the Apriori algorithm to explore the weighted relationship between the mastery degree of knowledge points and problem-solving characteristics to generate optimization suggestions for the knowledge point mastery path.

9. A computer device, characterized in that: include: A memory, a processor, and a computer program stored in the memory, which, when executed on the processor, implements the teaching optimization method based on knowledge space as claimed in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that A computer program is stored thereon, and when the computer program is executed by a processor, the teaching optimization method based on the knowledge space as claimed in any one of claims 1 to 7 is implemented.

Images