Individualized test paper generation and error question pushing teaching auxiliary system based on error question index card

Abstract

The application relates to the technical field of electronic question and answer teaching equipment, in particular to a personalized test paper setting and mistake question pushing teaching auxiliary system based on a mistake question index card, which comprises an electronic question and answer terminal and is used for executing the following: generating a mistake question index card after determining that a question is answered incorrectly; acquiring a knowledge node mistake question set belonging to the same knowledge point node; determining the node exposure degree of each knowledge point node and the learning process salience of a topological connection between adjacent knowledge point nodes; generating a candidate pushing path and determining the path tendency degree of each candidate pushing path; determining a target pushing path, and generating personalized test paper setting content in the order of the knowledge point nodes in the target pushing path. The application analyzes the learning gap corresponding to the mistake question through three levels of node exposure, connection abnormality and path generation, and converts mistake question pushing from single knowledge point supplement to progressive test paper setting and pushing along a knowledge path.

Description

Technical Field

[0001] This invention relates to the field of electronic question-and-answer teaching equipment technology, specifically to a personalized test paper generation and error-answer push teaching assistance system based on error-answer index cards. Background Technology

[0002] In electronic question-and-answer teaching devices, student answer data is typically collected following a process of question creation, answering, judgment, feedback, and correction. Error records are then generated based on question identifiers, knowledge point identifiers, question type identifiers, answering time, and correction evaluation results. To support personalized teaching, these error records are used to analyze students' mastery of each knowledge point and to construct learning paths by combining the relationships between knowledge points. This provides a structured basis for subsequent error-based question delivery and personalized test paper generation.

[0003] However, current technologies for processing incorrect question data mainly focus on the sequential relationships between knowledge points, making it difficult to further differentiate between the weak exposure of individual knowledge point nodes and the abnormal learning connections between adjacent knowledge point nodes within the knowledge point topology. Consequently, when generating incorrect question recommendations or personalized test papers, electronic question-and-answer devices tend to configure questions around individual weak knowledge points or general sequential relationships, making it difficult to determine whether to recommend basic reinforcement questions, adjacent knowledge point differentiation questions, or subsequent application questions. Furthermore, when multiple adjacent knowledge points show consecutive incorrect questions, it is difficult to establish a test paper composition path connecting prior knowledge to current knowledge to subsequent applications, resulting in insufficient targeting of incorrect question recommendations and personalized test paper composition. Summary of the Invention

[0004] This invention provides a personalized test paper generation and error question push teaching assistance system based on error question index cards to solve the above problems.

[0005] The personalized test paper generation and error-response teaching assistance system based on error index cards of the present invention adopts the following technical solution: One embodiment of the present invention provides a personalized test paper generation and error question push teaching assistance system based on error question index cards, including an electronic Q&A terminal. The electronic Q&A terminal is used to perform the following: after determining that a question was answered incorrectly, a mistake question index card is generated; wherein, the mistake question index card includes a mistake question attribution identifier; based on the mistake question attribution identifier, each mistake question index card is mapped to a corresponding knowledge point node in a knowledge point topology graph to obtain a set of mistake questions belonging to the same knowledge point node; based on the set of mistake questions for each knowledge point node, the node exposure degree of each knowledge point node and the learning process saliency of the topological connection between adjacent knowledge point nodes are determined; wherein, the node exposure degree is used to quantify the degree of insufficient knowledge point mastery reflected by the set of mistake questions for each knowledge point node, and the learning process saliency is used to quantify the degree of abnormality in the learning connection process between adjacent knowledge point nodes; based on the learning process saliency, candidate push paths are generated and the path tendency of each candidate push path is determined; wherein, the path tendency is used to quantify the degree of overall abnormality and continuous abnormal distribution of the candidate push paths; based on the path tendency, a target push path is determined, and personalized test paper generation content is generated along the sequential order of the knowledge point nodes in the target push path.

[0006] Furthermore, the method for determining the degree of node exposure includes: for each knowledge point node, extracting the effective answering time and standardized score of each incorrect question index card from the corresponding incorrect question set of the knowledge point node; determining the effective answering coefficient of the incorrect question index card based on the effective answering time and the reference answering time of the corresponding question type; determining the degree of insufficient mastery of the incorrect question index card based on the standardized score; determining the weakness exposure contribution of each incorrect question index card based on the degree of insufficient mastery; and determining the node exposure degree of each knowledge point node by comprehensively considering the weakness exposure contributions of each incorrect question index card belonging to the same knowledge point node.

[0007] Furthermore, the method for determining the saliency of the learning process includes: for each topological connection between adjacent knowledge point nodes, obtaining the node exposure degree of the knowledge point nodes at both ends of the topological connection; determining the node mastery stability degree of the knowledge point nodes at both ends based on the node exposure degree; determining the connection anomaly degree of the topological connection based on the node exposure degree; and determining the learning process saliency of the topological connection based on the connection anomaly degree and the node mastery stability degree of the knowledge point nodes at both ends.

[0008] Furthermore, determining the connection anomaly degree of the topology connection based on the node exposure degree includes: determining the common weakness degree and the mastery difference degree of the knowledge point nodes at both ends based on the node exposure degree; and determining the connection anomaly degree of the topology connection based on the common weakness degree and the mastery difference degree.

[0009] Furthermore, generating candidate push paths based on the saliency of the learning process includes: determining significant connections in the knowledge point topology graph based on the saliency of the learning process of each of the topology connections; for each significant connection, searching for adjacent topology connections along the preceding and following directions of the knowledge point topology graph; when an adjacent topology connection is found that satisfies the connection saliency condition, incorporating the adjacent topology connection and the new knowledge point node connected by the adjacent topology connection into the current candidate push path.

[0010] Furthermore, the step of generating candidate push paths based on the saliency of the learning process further includes: when the number of knowledge point nodes in the current candidate push path reaches the preset node number limit, and the next adjacent topological connection found by continuing to search along the current search direction satisfies the connection saliency condition, the end knowledge point node of the current candidate push path is taken as the new starting node, and the next candidate push path is reconstructed based on the next adjacent topological connection.

[0011] Further, determining the path tendency of each candidate push path includes: for each candidate push path, obtaining the learning process saliency of each topological connection in the candidate push path; determining the overall path anomaly level index and the path anomaly continuous fluctuation level index based on the learning process saliency; and determining the path tendency of the candidate push path based on the overall path anomaly level index and the path anomaly continuous fluctuation level index.

[0012] Further, the step of determining the target push path based on the path tendency and generating personalized test paper content along the order of the knowledge point nodes in the target push path includes: determining a pushable path among the candidate push paths; wherein, the pushable path is the candidate push path whose path tendency is not less than the path push threshold; sorting the pushable paths in descending order of path tendency, determining the pushable path at the top of the sort as the target push path, and the remaining pushable paths as paths to be pushed; and generating personalized test paper content along the order of the knowledge point nodes in the target push path.

[0013] Furthermore, generating personalized test content along the sequential order of the knowledge point nodes in the target push path includes: configuring basic reinforcement questions for the knowledge point nodes located at the starting position of the sequential order in the target push path; configuring transfer application questions for the knowledge point nodes located at the ending position of the sequential order; configuring consolidation questions for the knowledge point nodes located between the starting position and the ending position; and configuring questions based on the questions in the error index card corresponding to each knowledge point node. Furthermore, the method for obtaining the knowledge point topology graph includes: constructing the knowledge point topology graph based on the course syllabus, examination syllabus, textbook chapter structure, knowledge point annotations in the question bank, and the pre-configured knowledge point dependencies by the teacher; wherein, the knowledge point topology graph includes multiple knowledge point nodes and topological connections between knowledge point nodes.

[0014] The beneficial effects of the technical solution of the present invention are: In this embodiment of the invention, the personalized test paper generation and error-answering teaching assistance system based on error-answering index cards includes an electronic Q&A terminal. The electronic Q&A terminal is used to perform the following: after determining that a question was answered incorrectly, an error-answering index card is generated; wherein, the error-answering index card includes an error-answering identifier; based on the error-answering identifier, each error-answering index card is mapped to a corresponding knowledge point node in the knowledge point topology graph to obtain a set of error-answering questions belonging to the same knowledge point node; based on the set of error-answering questions for each knowledge point node, the node exposure degree of each knowledge point node and the learning process saliency of the topological connection between adjacent knowledge point nodes are determined; wherein, the node exposure degree is used to quantify the degree of insufficient knowledge point mastery reflected by the set of error-answering questions for each knowledge point node, and the learning process saliency is used to quantify the degree of abnormality in the learning connection process between adjacent knowledge point nodes; based on the learning process saliency, candidate push paths are generated and the path tendency of each candidate push path is determined; wherein, the path tendency is used to quantify the degree of overall abnormality and continuous abnormal distribution of the candidate push paths; based on the path tendency, a target push path is determined, and personalized test paper generation content is generated along the order of knowledge point nodes in the target push path.

[0015] This invention constructs a wrong-answer index card and maps it to a knowledge point topology map. It analyzes the learning gaps corresponding to wrong answers in three progressive levels: node exposure, connection anomalies, and path generation. It transforms the push of wrong answers from single knowledge point reinforcement to a progressive test paper generation and push along the knowledge path. On the other hand, by analyzing the degree of node exposure, it filters out valid answers and real evidence of insufficient mastery from the set of wrong answers for knowledge nodes, reducing the interference of guessing, skipping, or low-input answers on the judgment of weak nodes. Furthermore, by identifying two types of connection anomalies—shared weakness at both ends of adjacent knowledge point nodes and gaps in mastery at both ends—it accurately judges the problems in the process of knowledge connection, analysis, or transfer. Finally, by calculating path tendency and determining the target push path, it combines scattered significant connections into candidate paths that are abnormal as a whole and continuously distributed, and generates personalized test paper content in the order of prior knowledge to current knowledge to subsequent application. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the architecture of a personalized test paper generation and error-answering teaching assistance system based on error-answer index cards provided in an embodiment of the present invention. Figure 2 This is a flowchart illustrating the scheme executed by the electronic question-and-answer terminal provided in an embodiment of the present invention. Detailed Implementation

[0018] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following detailed description, in conjunction with the accompanying drawings and preferred embodiments, provides a detailed account of the specific implementation methods, structures, features, and effects of the method proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments may be combined in any suitable form. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] The following description, in conjunction with the accompanying drawings, details the specific solution of the personalized test paper generation and error-prone question push teaching assistance system based on error index cards provided by this invention.

[0020] like Figure 1 As shown, this embodiment of the invention provides a personalized test paper generation and error push teaching assistance system 100 based on error index cards, which includes an electronic question and answer terminal 110.

[0021] The aforementioned electronic question-and-answer terminal 110 can output question content to students and collect students' answers to the questions. The electronic question-and-answer terminal 110 can be implemented using a computing device with a display module and an input module, such as a personal computer, tablet, or smartphone. It presents question information to students in text, image, or multimedia format via a graphical user interface and receives students' answer input via a keyboard, touchscreen, mouse, or audio acquisition device. After a student submits their answer, the electronic question-and-answer terminal 110 can automatically judge the answer and present the judgment result and correction feedback to the student, thus completing a full interactive closed loop of question generation, answering, judgment, and feedback.

[0022] In some application scenarios, the aforementioned personalized test paper generation and error question push teaching assistance system 100 based on error question index cards is also equipped with a user terminal. This user terminal establishes a communication connection with the electronic question-and-answer terminal 110, which is used for teachers or teaching administrators to remotely input knowledge point topology map configuration information, preset values ​​for question type reference answering time, and push threshold parameters. The above configuration data is synchronized to the electronic question-and-answer terminal 110 to realize remote configuration and dynamic adjustment of knowledge point topology structure, node exposure degree calculation threshold, and personalized test paper generation rules. This allows teachers to intervene and optimize the push path filtering conditions and test paper question type ratio based on the distribution of students' error question data.

[0023] like Figure 2 As shown, the above-mentioned electronic question-and-answer terminal 110 is used to perform: Step S210: After determining that the question was answered incorrectly, generate an incorrect question index card; wherein, the incorrect question index card includes an identifier indicating the category of the incorrect question.

[0024] Students on the questions After the student answers the question, the electronic question-and-answer terminal 110 can automatically determine the student's answer. If the answer is incorrect, it will generate a corresponding error index card for that question. Each incorrect question corresponds to an incorrect question index card. Error Index Card Data on behaviors related to incorrect answers Correcting and reworking data and the identification of incorrect questions It consists of three parts, namely: .

[0025] Data on incorrect answers Including learning segment number and effective response time Two fields, including the learning segment number. According to the question The timing of incorrect responses is mapped according to a fixed learning segment length. One calendar day can be used as a learning segment, and when using one calendar day as a learning segment, the above learning segment number is... It can indicate the specific day the incorrect answer was given. Effective answering time. It can record the actual time a student spends answering a question from receiving it to submitting their answer. This parameter is used to reflect whether the wrong answer was generated during a valid question-and-answer process.

[0026] Correction and copying data The range of values ​​is Used to quantify this question The standardized scoring is as follows: 100 points for a correct answer and 0 points for an incorrect answer for objective questions; subjective questions are scored according to the conversion of the scores to a value between 0 and 100 points, and the corrected and repeated data is used. The sources include the scores of students who answer questions again and the system's or teachers' ratings of the corrected content.

[0027] Error Attribution Identifier Including knowledge point labels and question type identifiers Two fields, namely Among them, knowledge point identifiers Used to determine the question Corresponding knowledge points and question type identifiers Used to determine the question The corresponding question type category. The incorrect question attribution marker does not directly participate in numerical calculations; its core function is to classify incorrect questions... Error Index Card Mapped to the corresponding knowledge point nodes in the knowledge point topology graph .

[0028] Step S220: Based on the incorrect question attribution identifier, map each incorrect question index card to the corresponding knowledge point node in the knowledge point topology diagram to obtain the set of incorrect questions belonging to the same knowledge point node.

[0029] Step S220 above can map to the same knowledge point node based on the knowledge point identifier in the incorrect question attribution identifier. Multiple error index cards Merge the incorrect questions to form a set of incorrect questions for each knowledge point. .

[0030] Optionally, the method for obtaining the knowledge point topology graph includes: constructing a knowledge point topology graph based on the course syllabus, examination syllabus, textbook chapter structure, knowledge point annotations in the question bank, and the pre-configured knowledge point dependencies by the teacher; wherein, the knowledge point topology graph includes multiple knowledge point nodes and topological connections between knowledge point nodes.

[0031] The aforementioned course syllabus, examination syllabus, and textbook chapter structure together constitute the basic framework of the knowledge point topology diagram, used to determine the overall set of knowledge point nodes and their basic positional relationships within the subject system. Specifically, the course syllabus specifies the knowledge scope and teaching objectives of the course, the examination syllabus clarifies the assessment requirements and knowledge coverage, and the textbook chapter structure provides the arrangement order and hierarchical relationship of knowledge points.

[0032] The aforementioned knowledge point dependencies and annotation records are used to refine the connection weights and directions between knowledge point nodes on top of the basic framework. Specifically, the knowledge point annotation records for each question in the question bank record the correspondence between specific questions and knowledge points, reflecting the examination methods and distribution characteristics of knowledge points in actual assessments. The knowledge point dependencies pre-configured by teachers supplement the learning logic constraints between preceding and subsequent knowledge, such as the progressive relationship between basic concept nodes and comprehensive application nodes.

[0033] Based on the course syllabus, examination syllabus, textbook chapter structure, knowledge point annotations in the question bank, and the pre-configured knowledge point dependencies by teachers, a knowledge point topology graph is constructed by abstracting each independent knowledge point into a knowledge point node and the dependencies or relationships between knowledge points into topological connections. The knowledge point topology graph represents the path relationships between knowledge point nodes in a graph structure, where each knowledge point node corresponds to a target location that can be mapped to a wrong answer index card, and each topological connection corresponds to a learning connection path between adjacent knowledge point nodes.

[0034] Step S230: Based on the set of incorrect questions for each knowledge node, determine the node exposure degree of each knowledge node and the learning process salience of the topological connection between adjacent knowledge node nodes; wherein, the node exposure degree is used to quantify the degree of insufficient mastery of knowledge points reflected by the set of incorrect questions for each knowledge node, and the learning process salience is used to quantify the degree of abnormality in the learning connection process between adjacent knowledge node nodes.

[0035] It is understandable that a single incorrect question only records the error result in one question-and-answer process, which is insufficient to indicate that the corresponding knowledge point node has formed a stable weak state. Only when multiple incorrect question index cards are mapped and merged to the same knowledge point node through incorrect question attribution identifiers can these incorrect question data collectively reflect the abnormal mastery state of that node in the electronic question-and-answer process. This embodiment of the invention uses the node exposure degree to judge the weak exposure of individual knowledge point nodes in the knowledge point topology graph, in order to determine which nodes have genuine insufficient mastery, thereby providing node status input for subsequent analysis of learning connection anomalies between adjacent knowledge point nodes. The following describes one scheme for determining the node exposure degree: Optionally, the method for determining the degree of node exposure mentioned above includes: for each knowledge point node, extracting the effective answering time and standardized score of each incorrect question index card from the corresponding knowledge point node incorrect question set; determining the effective answering coefficient of the incorrect question index card based on the effective answering time and the reference answering time of the corresponding question type; determining the degree of insufficient mastery of the incorrect question index card based on the standardized score; determining the weakness exposure contribution of each incorrect question index card based on the degree of insufficient mastery; and determining the node exposure degree of each knowledge point node by comprehensively considering the weakness exposure contributions of each incorrect question index card belonging to the same knowledge point node.

[0036] For each incorrect question index card in the knowledge node incorrect question set, extract the effective answering time recorded therein. Suggested answering time for the corresponding question type Reference time for answering The effective response time is determined by the average historical response time for the same question type in the question bank. If historical data is insufficient, it is preset by the teacher or question bank configuration terminal based on the question type difficulty. Calculate the effective response time. Suggested answering time for the corresponding question type The ratio of the two values, and the smaller of the ratio and the value of 1, yields the effective response coefficient. The effective response coefficient ranges from 0 to 1. When the actual response time reaches or exceeds the reference response time, the value is 1. When the actual response time is less than the reference response time, the value is the actual ratio.

[0037] Extract standardized scores from the same error index card. Standardized scores After dividing by 100, take the difference between 1 and the ratio. As a measure of the degree of mastery deficiency, the degree of mastery deficiency maps the original score to a coefficient of non-achievement between 0 and 1. The higher the value, the greater the degree of mastery deficiency for that incorrect question under the current evaluation result. The value approaches 0 when the student has completely corrected and answered the question correctly again, and the value is 1 when the answer is completely wrong.

[0038] Multiplying the above effective response coefficient by the degree of insufficient mastery yields the contribution of the error index card to the exposure of weaknesses in the corresponding knowledge point node, namely: As can be understood from the calculation formula for the contribution of the weakness exposure of knowledge point nodes mentioned above, only when the wrong answer has an effective answering process and the standardized score is low will it have a high contribution to the degree of weakness exposure of the knowledge point node; if the student answers wrongly without effective answering, or if the wrong answer has been corrected and digested, the product result will be lower accordingly. Therefore, this product filters out the effective wrong answer evidence that can truly reflect the current weakness of the node from the set of wrong answers for the knowledge point node, avoiding the misinterpretation of guessed answers, skipped answers, low-intensity answers, or wrong answers that have been corrected and digested as the true basis for weakness.

[0039] For nodes belonging to the same knowledge point The arithmetic mean of the weaknesses exposed by all the error index cards is used to obtain the node exposure degree of that knowledge point node. The calculation formula is:

[0040] In the formula, , representing knowledge point nodes The number of incorrect questions (or knowledge points) (The corresponding number of incorrect question index cards); Representing knowledge point nodes A collection of incorrect answers related to knowledge nodes. Node exposure level. The value ranges from 0 to 1. The larger the value, the higher the degree of weakness of the knowledge point node in the electronic question and answer process. The smaller the value, the more stable the mastery of the node.

[0041] It should be noted that the above scheme is only one way to determine the degree of node exposure, and not the only way. The embodiments of the present invention can also use one of the following schemes to determine the degree of node exposure: (1) Determine it directly based on the ratio of the number of wrong questions in the set of wrong questions of the knowledge node to the total number of answers under the knowledge node, so as to reflect the degree of insufficient mastery with error frequency. (2) According to the standardized score change trend in the multiple answering trajectory under the same knowledge node, when the score of multiple consecutive answers is continuously lower than the preset score, the exposure degree of the knowledge node is increased accordingly to identify the persistent weak nodes that recur.

[0042] It is understandable that in the actual use of electronic question-and-answer teaching devices, a single incorrect question is often not an isolated knowledge point error, but may reflect a weakness in a certain knowledge point node, an abnormal connection between adjacent knowledge points, or a break in the continuous learning path formed by connecting multiple knowledge points. Simply relying on the degree of weakness exposure of a single knowledge point node cannot distinguish whether there is a continuity break between adjacent knowledge point nodes due to shared weaknesses, nor can it identify the gap in mastery between the two ends of the node reflecting a poor connection between prior knowledge and current knowledge, a failure to transfer current knowledge to subsequent application question types, or unclear distinction between adjacent easily confused knowledge points. Therefore, this embodiment of the invention, after determining the node exposure degree of each knowledge point node, can further determine the salience of the learning process of the topological connection between adjacent knowledge point nodes, so as to transfer the incorrect question data from the node level to the connection level, determine whether there are abnormalities in the learning connection process between adjacent knowledge points, and thus provide a connection-level analytical basis for subsequently determining whether significant connections can form a continuous push path. The following describes a scheme for determining the salience of the learning process: Optionally, the method for determining the saliency of the learning process includes: for each topological connection between adjacent knowledge point nodes, obtaining the node exposure degree of the knowledge point nodes at both ends of the topological connection; determining the node mastery stability of the knowledge point nodes at both ends based on the node exposure degree; determining the connection anomaly degree of the topological connection based on the node exposure degree; and determining the learning process saliency of the topological connection based on the connection anomaly degree and the node mastery stability of the knowledge point nodes at both ends.

[0043] For node exposure level The knowledge points and nodes, and the stability of their mastery. The value is 1 and The difference, that is Due to the degree of node exposure The value ranges from 0 to 1, and the node controls the stability. The value range is also from 0 to 1. A higher node exposure indicates a greater weakness in that knowledge point, and consequently, a lower level of stability in mastering that knowledge point; conversely, a lower node exposure indicates a more stable mastery of that knowledge point. For knowledge point nodes... and knowledge point nodes The node exposure level was obtained respectively. and and the stability of node control and Knowledge Points and The connection between them in the knowledge point topology diagram is denoted as .

[0044] Optionally, the above-mentioned determination of the connection anomaly degree of the topology connection based on the node exposure degree includes: determining the common weakness degree and the mastery difference degree of the knowledge point nodes at both ends based on the node exposure degree; and determining the connection anomaly degree of the topology connection based on the common weakness degree and the mastery difference degree.

[0045] Calculate the topology connections based on the node exposure levels of the knowledge point nodes at both ends. abnormal connection strength The calculation formula is as follows:

[0046] In the formula, Representing knowledge point nodes and knowledge point nodes The common vulnerability level only increases when both nodes have high vulnerability exposure. This indicates the degree of difference in mastery status between two nodes, reflecting whether there is a significant gap in mastery between the two ends. Therefore, connection anomaly strength... This study comprehensively characterizes two types of connection anomalies: common weakness of both nodes and discontinuous control of the two nodes' states.

[0047] Furthermore, based on the connection anomaly strength and the node mastery stability of the two endpoints, the salience of the learning process is calculated. The calculation formula is:

[0048] In the formula, For smoothing parameters, used to avoid When the denominator is zero, its value can be 0.01. This relates to the significance of the above learning process. In the calculation formula, This represents the correction result of the connection anomaly strength relative to the stability of the two end nodes; when the connection anomaly strength is large and the stability of both end nodes is low, this ratio increases; when both end nodes are relatively stable, i.e. When it is large, It is suppressed.

[0049] Significance of the learning process Used to characterize topological connections The higher the value of the abnormal significance level under the current incorrect question data, the more the knowledge connection, knowledge identification or knowledge transfer process corresponding to the topological connection should be included in the subsequent candidate push path; the lower the value, the lower the push reference value of the topological connection under the current incorrect question data.

[0050] It should be noted that the above scheme is only one way to determine the saliency of the learning process, and not the only way. The embodiments of the present invention can also use one of the following schemes to determine the saliency of the learning process: (1) directly determine it by the geometric mean of the node exposure degree of the knowledge point nodes at both ends, and use the common weakness degree as a measure of connection anomaly. (2) use the ratio of the absolute value of the difference between the node exposure degree of the two ends to the sum of the two, that is, the relative difference coefficient, to quantify the connection anomaly caused by the mastery gap.

[0051] Step S240: Based on the saliency of the learning process, generate candidate push paths and determine the path tendency of each candidate push path; wherein, the path tendency is used to quantify the degree of overall abnormality and continuous distribution of abnormality of the candidate push paths.

[0052] Optionally, the above-mentioned generation of candidate push paths based on the saliency of the learning process includes: determining significant connections in the knowledge point topology graph based on the saliency of the learning process of each topology connection; for each significant connection, searching for adjacent topology connections along the preceding and following directions of the knowledge point topology graph respectively; when an adjacent topology connection is found that satisfies the connection saliency condition, merging the adjacent topology connection and the new knowledge point node connected by the adjacent topology connection into the current candidate push path.

[0053] Based on the saliency of the learning process of each topological connection, significant connections are selected from the knowledge point topology graph. The criterion for a significant connection is that the saliency of the learning process is not less than the significance threshold of the connection, that is: Significant threshold for connecting lines In the formula, This represents the arithmetic mean of the salience of the learning process for all topological connections in the topological graph for the same knowledge point. The standard deviation is the corresponding value. The significance threshold for the above connection can be adaptively determined based on the distribution of learning process anomalies on each current topology connection. Topology connections with learning process significance higher than the threshold are identified as significant connections, while topology connections with learning process significance lower than the threshold are not included in the candidate push path.

[0054] For each significant connection, adjacent topological connections are searched along the preceding and following directions of the knowledge point topology graph. The preceding direction is the direction from the current knowledge point node to its preceding concept node, basic rule node, or preceding method node, and the following direction is the direction from the current knowledge point node to its subsequent application node, comprehensive question type node, or transfer application node. In either search direction, the learning process saliency of the topological connections between the current knowledge point node and its adjacent knowledge point nodes is read sequentially. When an adjacent topological connection meets the connection saliency condition, the adjacent topological connection and the new knowledge point node connected by the adjacent topological connection are incorporated into the current candidate push path.

[0055] Optionally, the above-mentioned generation of candidate push paths based on the saliency of the learning process further includes: when the number of knowledge point nodes in the current candidate push path reaches the preset node limit, and the next adjacent topological connection found along the current search direction satisfies the connection saliency condition, the end knowledge point node of the current candidate push path is taken as the new starting node, and the next candidate push path is reconstructed based on the next adjacent topological connection.

[0056] During the generation of candidate push paths, an upper limit threshold for the number of knowledge point nodes is set to control the knowledge coverage of a single candidate push path. In some application scenarios, this upper limit threshold is set to 4 knowledge point nodes, corresponding to 3 topology connections. Assume the current candidate path is: In the formula, This indicates the number of topological connections included in the candidate push path. Therefore, the above constraint on the number of knowledge points satisfies: , ,in, Representing a path The number of knowledge point nodes included. Representing a path The number of connections included. Understandably, limiting the number of knowledge point nodes in the candidate push path aims to avoid covering an excessively long knowledge chain in a single push, preventing the cognitive load on learners from exceeding a reasonable range due to an excessive number of knowledge point nodes, while ensuring that each candidate push path focuses on a manageable continuous learning anomaly interval, so that the personalized test content generated subsequently can accurately supplement the limited knowledge interval.

[0057] When the number of knowledge point nodes in the current candidate push path has reached the aforementioned upper limit threshold, and the next adjacent topological connection found along the current search direction still satisfies the connection salience condition, a path sliding splitting mechanism is triggered. This mechanism uses the end knowledge point node of the current candidate push path as the new starting node, and reconstructs the next candidate push path based on this next adjacent topological connection. The end knowledge point node of the original candidate push path serves as both the end point of the previous path and the starting point of the next path, achieving overlapping connection between the two candidate push paths.

[0058] Through the above search and splitting process, a set of candidate push paths is obtained, consisting of continuous, interconnected topological connections that satisfy the saliency condition. In the formula, This represents the number of candidate push paths. Each candidate push path corresponds to a learning connection chain with learning process anomalies in the knowledge point topology graph, providing the path object basis for subsequent calculation of path preference and determination of the target push path.

[0059] It is understandable that the actual learning process usually unfolds continuously in the order of prior knowledge to current knowledge and then to subsequent application. A single significant connection can only reflect a local learning connection anomaly between two adjacent knowledge point nodes, and is insufficient to determine whether the anomaly exists in isolation or extends further along the knowledge chain in the direction of the preceding or subsequent knowledge points. When multiple significant connections are continuously distributed in the knowledge point topology graph, it may indicate a break in the continuous learning path from basic concepts to comprehensive applications. In this case, configuring local reinforcement questions around only one significant connection will not be able to cover the complete knowledge gap. Therefore, after determining the learning process significance of each topology connection, the path tendency of the candidate push path can be further calculated to measure the degree of overall significant and continuous distribution of the path anomaly. This allows us to determine which continuous paths composed of significant connections have overall push value, and based on this, determine the target push path suitable for personalized test paper generation and incorrect question push. The following introduces a scheme for determining path tendency: Optionally, the above determination of the path tendency of each candidate push path includes: for each candidate push path, obtaining the learning process saliency of each topological connection in the candidate push path; based on the learning process saliency, determining the overall path anomaly level index and the path anomaly continuous fluctuation level index; and based on the overall anomaly level index and the path anomaly continuous fluctuation level index, determining the path tendency of the candidate push path.

[0060] For any candidate push path, the saliency of the learning process of each topological connection in that path is read. A candidate push path consists of consecutive, connected topological connections that satisfy the connection saliency condition, denoted as […]. , Indicates the first in the path The learning process saliency of all topological connections in the path. The arithmetic mean of the learning process saliencies of all topological connections in the path is calculated and used as an indicator of the overall anomaly level of the path. The calculation formula is as follows: The overall path anomaly level index reflects the overall significance of anomalies in candidate push paths. When the learning process significance of each topological connection in the path is high, the index increases accordingly, indicating that the path as a whole has obvious learning process anomalies and has high basic push value.

[0061] The relative change in saliency between two adjacent topological connections during the learning process is calculated as an indicator of the level of continuous fluctuation in path anomalies. For adjacent... Article and No. The topological connections are used to calculate the significance of the two learning processes and the smoothing parameter, respectively. The ratio of the sums, taken as the natural logarithm and then the absolute value, is... The arithmetic mean of the changes in the above values ​​for all adjacent line pairs is then added to obtain the path anomaly continuous fluctuation level index. Among them, the smoothing parameter To prevent the logarithmic term from being uncalculated when the significance of the learning process is 0, the value can be 0.01. The path anomaly continuity fluctuation level index reflects the continuity of the abnormal distribution in the candidate push path. When the significance of adjacent lines jumps significantly, this index increases accordingly, thus suppressing paths with discontinuous significance.

[0062] Dividing the overall path anomaly level index by the path anomaly continuous fluctuation level index yields the path tendency of the candidate push path. The calculation formula is:

[0063] When the candidate push path contains only one topology connection, that is When adjacent connection pairs do not exist, and there is no basis for calculating the level index of abnormal continuous fluctuations in the path, the learning process significance of this topological connection is directly used as the path tendency of the candidate push path. .

[0064] The aforementioned path tendency takes into account both the overall significance level and the degree of continuous distribution of anomalies in the candidate push paths. The higher the overall anomaly level index of the path, the greater the path tendency, indicating that there are obvious learning connection anomalies in the path as a whole. The higher the continuous fluctuation level index of the path anomalies, the smaller the path tendency, which suppresses candidate paths with discontinuous significance and obvious jumps.

[0065] It should be noted that the above scheme is only one way to determine path preference, and not the only way. Embodiments of the present invention can also use one of the following schemes to determine path preference: (1) Directly take the arithmetic mean of the significance of the learning process of each topology connection in the candidate push path, and use the overall abnormality level of the path as the basis for ranking push priorities. (2) Use the minimum value of the significance of the learning process of each topology connection, and use the weakest connection link in the path as the lower limit measure of the path push value.

[0066] Step S250: Determine the target push path based on path preference, and generate personalized test paper content along the order of knowledge point nodes in the target push path.

[0067] Unlike the traditional fixed test paper format, the personalized test paper generation refers to the process of dynamically configuring and generating test paper content targeting the student's current weaknesses based on the distribution of their incorrect answers, their mastery of knowledge points, and the characteristics of their learning path. By mapping the incorrect answer data generated by the student during the electronic Q&A process to the knowledge point topology, identifying weak points and path breakpoints, and then selectively configuring questions along the order of knowledge point nodes in the target push path, the generated test paper content matches the student's actual learning gaps in terms of knowledge point coverage and difficulty progression.

[0068] Optionally, step S250 includes: determining a pushable path from each candidate push path; wherein, a pushable path is a candidate push path with a path inclination degree not less than the path push threshold; sorting each pushable path in descending order of path inclination degree, determining the pushable path at the top of the sort as the target push path, and the remaining pushable paths as paths to be pushed; generating personalized test paper content along the order of knowledge point nodes in the target push path.

[0069] The path preference of each candidate push path is compared with the path push threshold. The path push threshold can be adaptively determined by the path preference distribution of all current candidate push paths, and the calculation formula is as follows: In the formula, This is the arithmetic mean of the path preference for each candidate push path. The standard deviation corresponds to the path tendency. When the path tendency of a candidate push path is not less than the push threshold for that path, the candidate push path is determined to be a pushable path, indicating that it has both overall saliency and continuous stability, and can be used as the path-based push target for the current student. When the path tendency of a candidate push path is less than the push threshold for that path, it indicates that the overall abnormality of the path is insufficient or the path continuity is insufficient. In this case, complete test content will not be generated along the path, and only abnormal nodes or abnormal connections in the path will be retained as a basis for subsequent observation or local review.

[0070] Optionally, the above-mentioned generation of personalized test paper content along the sequential order of knowledge point nodes in the target push path includes: configuring basic reinforcement questions for knowledge point nodes located at the starting position of the sequential order in the target push path; configuring transfer application questions for knowledge point nodes located at the ending position of the sequential order; configuring consolidation questions for knowledge point nodes located between the starting and ending positions; and determining the question type of the questions configured for the corresponding knowledge point node based on the question type identifier in the error index card corresponding to each knowledge point node.

[0071] When generating personalized test content based on the sequential order of knowledge point nodes along the target push path, different levels of questions can be configured according to the relative position of each knowledge point node in the path. For example, a knowledge point node at the starting point of the sequence corresponds to prerequisite knowledge, and basic reinforcement questions are configured for this node. These questions focus on reviewing prerequisite concepts, basic rules, or prerequisite methods to correct any deficiencies in mastery exposed during the electronic Q&A process. A knowledge point node at the ending point of the sequence corresponds to subsequent application knowledge, and transfer application questions are configured for this node. These questions focus on expanding the current knowledge to comprehensive question types or transfer application scenarios to test and correct any connection anomalies in the knowledge transfer process. A knowledge point node between the starting and ending points corresponds to the current core knowledge, and consolidation questions are configured for this node. These questions focus on strengthening the current knowledge point itself and may include variations of the same question type to consolidate the basic mastery of the node and enhance proficiency in the same question type.

[0072] By reading the question type identifiers from the error index cards under each knowledge point node, the specific question types configured for each knowledge point node are determined. These question type identifiers record the original question type category corresponding to the student's incorrect answers at that knowledge point node. Based on these identifiers, questions matching the incorrect question type are configured for the corresponding knowledge point node. For example, if a student's incorrect answers at a node are concentrated in multiple-choice questions, then reinforcement or application questions in the form of multiple-choice questions are configured; if incorrect answers are concentrated in open-ended questions, then reinforcement or application questions in the form of open-ended questions are configured. In this way, the test content not only matches the progressive relationship of knowledge point paths but also further matches the student's weak points in question types at each node.

[0073] Through the dual configuration mechanism of node positions and question types mentioned above, personalized test paper content forms a progressive structure along the target push path, from basic reinforcement to consolidation and reinforcement, and then to application and transfer. Basic reinforcement questions fill in gaps in prior knowledge, consolidation questions strengthen the mastery of current knowledge, and application and transfer questions connect with subsequent knowledge. The three unfold sequentially along the knowledge path, so that the push of incorrect questions no longer focuses on isolated reinforcement of single knowledge points, but is transformed into a layered and progressive test paper composition and push along a continuous knowledge path.

[0074] This invention is now complete.

[0075] In summary, in this embodiment of the invention, the personalized test paper generation and error-answering teaching assistance system based on error-answering index cards includes an electronic Q&A terminal. The electronic Q&A terminal is used to perform the following: after determining that a question was answered incorrectly, a problem-answering index card is generated; wherein, the problem-answering index card includes a problem-answering identifier; based on the problem-answering identifier, each problem-answering index card is mapped to a corresponding knowledge point node in the knowledge point topology graph to obtain a set of problem-answering questions belonging to the same knowledge point node; based on the set of problem-answering questions for each knowledge point node, the node exposure degree of each knowledge point node and the learning process saliency of the topological connection between adjacent knowledge point nodes are determined; wherein, the node exposure degree is used to quantify the degree of insufficient knowledge point mastery reflected by the set of problem-answering questions for each knowledge point node, and the learning process saliency is used to quantify the degree of abnormality in the learning connection process between adjacent knowledge point nodes; based on the learning process saliency, candidate push paths are generated and the path tendency of each candidate push path is determined; wherein, the path tendency is used to quantify the degree of overall abnormality and continuous abnormal distribution of the candidate push paths; based on the path tendency, a target push path is determined, and personalized test paper generation content is generated along the order of knowledge point nodes in the target push path.

[0076] This invention constructs a wrong-answer index card and maps it to a knowledge point topology map. It analyzes the learning gaps corresponding to wrong answers in a three-level progressive manner: node exposure, connection anomalies, and path generation. It transforms the push of wrong answers from single knowledge point reinforcement to a progressive test paper generation and push along the knowledge path. On the other hand, by analyzing the degree of node exposure, it filters out valid answers and real evidence of insufficient mastery from the set of wrong answers for knowledge nodes, reducing the interference of guessing, skipping, or low-input answers on the judgment of weak nodes. Furthermore, by identifying two types of connection anomalies, namely, common weakness at both ends of adjacent knowledge point nodes and gap in mastery at both ends, it accurately judges the problems in the knowledge connection, analysis, or transfer process. Finally, by calculating path tendency and determining the target push path, it combines scattered significant connections into candidate paths that are abnormal as a whole and continuously distributed, and generates personalized test paper content in the order of prior knowledge to current knowledge to subsequent application.

[0077] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A personalized test paper generation and error-answer push teaching assistance system based on error-answer index cards, characterized in that, Includes an electronic question-and-answer terminal, which is used to perform: After determining that a question was answered incorrectly, an incorrect question index card is generated; wherein, the incorrect question index card includes an identifier for the incorrect question; Based on the incorrect question attribution identifier, each incorrect question index card is mapped to the corresponding knowledge point node in the knowledge point topology diagram to obtain the knowledge node incorrect question set belonging to the same knowledge point node; Based on the set of incorrect answers for each knowledge node, the node exposure degree of each knowledge node and the salience of the learning process of the topological connection between adjacent knowledge node nodes are determined; wherein, the node exposure degree is used to quantify the degree of insufficient mastery of knowledge points reflected by the set of incorrect answers for each knowledge node, and the salience of the learning process is used to quantify the degree of abnormality in the learning connection process between adjacent knowledge node nodes. Based on the saliency of the learning process, candidate push paths are generated and the path tendency of each candidate push path is determined; wherein, the path tendency is used to quantify the degree to which the candidate push paths are generally abnormal and the abnormality is continuously distributed. The target push path is determined based on the path preference, and personalized test paper content is generated along the order of the knowledge point nodes in the target push path. The method for determining the saliency of the learning process includes: for each topological connection between adjacent knowledge point nodes, obtaining the node exposure degree of the knowledge point nodes at both ends of the topological connection; determining the node mastery stability degree of the knowledge point nodes at both ends based on the node exposure degree; determining the connection anomaly degree of the topological connection based on the node exposure degree; and determining the learning process saliency of the topological connection based on the connection anomaly degree and the node mastery stability degree of the knowledge point nodes at both ends. The step of generating candidate push paths based on the saliency of the learning process includes: determining significant connections in the knowledge point topology graph based on the saliency of the learning process of each of the topology connections; for each significant connection, searching for adjacent topology connections along the preceding and following directions of the knowledge point topology graph; when an adjacent topology connection is found that satisfies the connection saliency condition, incorporating the adjacent topology connection and the new knowledge point node connected by the adjacent topology connection into the current candidate push path.

2. The personalized test paper generation and error-answer push teaching assistance system based on error-answer index cards according to claim 1, characterized in that, The method for determining the node exposure level includes: For each knowledge point node, extract the effective answering time and standardized score of each incorrect question index card from the corresponding incorrect question set of the knowledge point node. Based on the effective answering time and the reference answering time for the corresponding question type, the effective answering coefficient of the incorrect question index card is determined; Based on the standardized scores, the degree of insufficient mastery of the incorrect question index cards is determined; Based on the degree of lack of mastery, the contribution of each of the error index cards to the exposure of weaknesses is determined; By comprehensively considering the weakness exposure contributions of each of the incorrect question index cards belonging to the same knowledge point node, the node exposure degree of each knowledge point node is determined.

3. The personalized test paper generation and error-answer push teaching assistance system based on error-answer index cards according to claim 1, characterized in that, The determination of the connection anomaly level of the topology based on the node exposure level includes: Based on the degree of node exposure, determine the common weakness and the degree of difference in mastery of the knowledge points at both ends; Based on the common weakness and the degree of mastery difference, the degree of connection anomaly of the topology is determined.

4. The personalized test paper generation and error-answer push teaching assistance system based on error-answer index cards according to claim 1, characterized in that, The step of generating candidate push paths based on the saliency of the learning process further includes: When the number of knowledge point nodes in the current candidate push path reaches the preset node limit, and the next adjacent topology connection found by continuing the search along the current search direction satisfies the connection significance condition, the end knowledge point node of the current candidate push path is taken as the new starting node, and the next candidate push path is reconstructed based on the next adjacent topology connection.

5. The personalized test paper generation and error-answer push teaching assistance system based on error-answer index cards according to claim 1, characterized in that, Determining the path preference of each of the candidate push paths includes: For each candidate push path, the learning process saliency of each topological connection in the candidate push path is obtained; Based on the saliency of the learning process, the overall path anomaly level index and the path anomaly continuous fluctuation level index are determined. Based on the overall anomaly level index and the path anomaly continuous fluctuation level index, the path tendency of the candidate push path is determined.

6. The personalized test paper generation and error-answer push teaching assistance system based on error-answer index cards according to any one of claims 1-5, characterized in that, The step of determining the target push path based on the path preference, and generating personalized test paper content along the sequential order of the knowledge point nodes in the target push path, includes: Among the candidate push paths, a pushable path is determined; wherein, the pushable path is the candidate push path whose path inclination is not less than the path push threshold; The pushable paths are sorted in descending order of path preference, and the pushable path at the top of the list is determined as the target push path, while the remaining pushable paths are designated as paths to be pushed. Personalized test paper content is generated according to the order of the knowledge point nodes in the target push path.

7. The personalized test paper generation and error-answer push teaching assistance system based on error-answer index cards according to claim 6, characterized in that, The process of generating personalized test paper content along the sequential order of the knowledge point nodes in the target push path includes: Configure basic reinforcement questions for the knowledge point nodes located at the starting position of the sequence in the target push path; Configure application problems for knowledge point nodes located at the end of the sequence; Configure reinforcement questions for the knowledge point nodes located between the starting position and the ending position; Based on the question type identifier in the error index card corresponding to each knowledge point node, the question type of the questions configured for the corresponding knowledge point node is determined.

8. The personalized test paper generation and error-answer push teaching assistance system based on error-answer index cards according to any one of claims 1-5, characterized in that, The method for obtaining the knowledge point topology graph includes: Based on the course syllabus, examination syllabus, textbook chapter structure, knowledge point annotations in the question bank, and the pre-configured knowledge point dependencies by the teacher, a knowledge point topology graph is constructed; wherein, the knowledge point topology graph includes multiple knowledge point nodes and topological connections between knowledge point nodes.

Images