Knowledge tracing method for applied learning transfer

Abstract

The present invention discloses a knowledge tracking method for applied learning transfer, which relates to the field of knowledge tracking processing. The method includes: forming a historical mathematics knowledge point set based on historical learning logs, analyzing and obtaining an initial mathematics knowledge point map; updating the real-time mathematics knowledge point set to the initial mathematics knowledge point map; obtaining a first mathematics knowledge point in the real-time mathematics knowledge point map, and obtaining a target knowledge tracking record; reading a predetermined tracking feature, and performing feature matching on the first tracking record; using a first predetermined coefficient as a weight, adjusting and calculating the first mastery degree obtained by weighting the variation of the first tracking feature parameter to obtain a target mastery degree. The method solves the technical problem that the existing knowledge tracking method is difficult to achieve effective transfer and sharing between different courses or knowledge points, resulting in low tracking and evaluation accuracy, and achieves the technical effect of improving knowledge tracking and evaluation accuracy by achieving effective transfer and sharing between different courses or knowledge points.

Description

Technical Field

[0001] The present application relates to the field of knowledge tracking and processing, and in particular to a knowledge tracking method for applying learning transfer. Background Art

[0002] With the rapid development of online education and intelligent learning systems, effectively tracking and evaluating students' learning progress and knowledge mastery has become a key issue. Knowledge tracking technology analyzes students' learning behavior and historical data to predict and assess their mastery of specific knowledge points. However, traditional knowledge tracking methods primarily focus on tracking knowledge points within a single course, assessing their mastery by analyzing their learning behavior within that course (such as their responses and study time). However, this approach cannot achieve the transfer and sharing of knowledge points between different courses.

[0003] Among the current related technologies, knowledge tracking methods have the technical problem of being difficult to achieve effective transfer and sharing between different courses or knowledge points, resulting in low accuracy of tracking and evaluation. Summary of the Invention

[0004] The present application provides a knowledge tracking method for applied learning transfer, which adopts the method of forming a historical mathematics knowledge point set based on the historical learning log of the target user, analyzing and obtaining an initial mathematics knowledge point map, updating the real-time mathematics knowledge point set obtained by decomposing the target course to the initial mathematics knowledge point map, obtaining the real-time mathematics knowledge point map, obtaining the first mathematics knowledge point in the real-time mathematics knowledge point map with the real-time mathematics knowledge point set as a constraint, and obtaining the target knowledge tracking record of the target user for the first mathematics knowledge point, reading the predetermined tracking feature, and performing feature matching on the first tracking record based on the predetermined tracking feature to obtain the first tracking feature parameter, obtaining the first predetermined coefficient, and using the first predetermined coefficient as a weight to adjust and calculate the first mastery degree obtained by weighting the variation of the first tracking feature parameter to obtain the target mastery degree and other technical means, thereby achieving the technical effect of improving the accuracy of knowledge tracking and evaluation by realizing effective migration and sharing between different courses or knowledge points.

[0005] This application provides a knowledge tracking method for applied learning transfer, including:

[0006] A historical mathematics knowledge point set is formed according to the historical learning log of the target user, and the historical mathematics knowledge point set is analyzed to obtain an initial mathematics knowledge point map; the real-time mathematics knowledge point set obtained by decomposing the target course is updated to the initial mathematics knowledge point map to obtain a real-time mathematics knowledge point map; a first mathematics knowledge point in the real-time mathematics knowledge point map is obtained with the real-time mathematics knowledge point set as a constraint, and a target knowledge tracking record of the target user for the first mathematics knowledge point is obtained, wherein the target knowledge tracking record includes a first tracking record; a predetermined tracking feature is read, and a feature matching is performed on the first tracking record based on the predetermined tracking feature to obtain a first tracking feature parameter; a first predetermined coefficient is obtained, and the first predetermined coefficient is used as a weight to adjust and calculate the first mastery degree obtained by weighting the variation of the first tracking feature parameter to obtain a target mastery degree, wherein the target mastery degree is used to characterize the mastery degree of the target user for the first mathematics knowledge point.

[0007] In a possible implementation, a historical mathematics knowledge point set is formed based on the historical learning log of the target user, and the historical mathematics knowledge point set is analyzed to obtain an initial mathematics knowledge point map, and the following processing is performed:

[0008] A task database is established, the task database including a plurality of tasks with mathematical knowledge point labels; a first historical mathematical knowledge point and a second historical mathematical knowledge point are randomly obtained from the set of historical mathematical knowledge points; a determination is made as to whether the first historical mathematical knowledge point and the second historical mathematical knowledge point co-occur in the first mathematical knowledge point label; if so, a reverse match is performed on the first task corresponding to the first mathematical knowledge point label, and the first task is added to a co-occurring task list; a ratio of the number of tasks in the co-occurring task list to the number of tasks in the task database is used as a correlation between the first historical mathematical knowledge point and the second historical mathematical knowledge point; and an initial mathematical knowledge point map is constructed by combining the first historical mathematical knowledge point and the second historical mathematical knowledge point with the correlation as a constraint.

[0009] In a possible implementation, the following processing is performed using the ratio of the number of tasks in the co-occurring task list to the number of tasks in the task database as the relevance between the first historical mathematics knowledge point and the second historical mathematics knowledge point:

[0010] Obtain a first history vector of the first historical mathematical knowledge point; obtain a second history vector of the second historical mathematical knowledge point; obtain similarity between the first history vector and the second history vector using a cosine similarity principle; and adjust the correlation using the similarity as a weight.

[0011] In a possible implementation, a first mathematical knowledge point in the real-time mathematical knowledge point graph is obtained with the real-time mathematical knowledge point set as a constraint, and a target knowledge tracking record of the target user for the first mathematical knowledge point is obtained, and the following processing is performed:

[0012] Extracting a second task from the multiple tasks with mathematical knowledge point tags, where the second task corresponds to a second mathematical knowledge point tag; determining whether the first mathematical knowledge point belongs to the second mathematical knowledge point tag; if so, adding the second task to a candidate task list; and monitoring a process in which the target user handles each task in the candidate task list to obtain the target knowledge tracking record.

[0013] In a possible implementation, the process of the target user processing each task in the candidate task list is monitored to obtain the target knowledge tracking record, and the following processing is performed:

[0014] The real-time mathematical knowledge point map is analyzed with the first mathematical knowledge point as the center and a predetermined distance as the radius to determine a first associated mathematical knowledge point range of the first mathematical knowledge point; any associated mathematical knowledge point in the first associated mathematical knowledge point range is extracted; and it is determined whether the any associated mathematical knowledge point belongs to a third-related mathematical knowledge point tag; if so, a third task corresponding to the third-related mathematical knowledge point tag is added to the candidate task list.

[0015] In a possible implementation, the following processing is performed:

[0016] The predetermined tracking features include task accuracy and unit task time.

[0017] In a possible implementation, a first predetermined coefficient is obtained, and the first predetermined coefficient is used as a weight to adjust and calculate the first mastery degree obtained by weighting the variation of the first tracking feature parameter to obtain a target mastery degree, and the following processing is performed:

[0018] Reversely match the first candidate task corresponding to the first tracking record, wherein the first candidate task corresponds to a first candidate mathematical knowledge point tag; randomly extract any candidate mathematical knowledge point in the first candidate mathematical knowledge point tag; obtain any correlation between the any candidate mathematical knowledge point and the first mathematical knowledge point in combination with the real-time mathematical knowledge point graph; take the average of the any correlations as the first task migration index of the first candidate task, and record the first task migration index as the first predetermined coefficient.

[0019] In a possible implementation, after obtaining any correlation between any candidate mathematical knowledge point and the first mathematical knowledge point in combination with the real-time mathematical knowledge point graph, the following processing is performed:

[0020] Obtain the learning time of any candidate mathematical knowledge point; analyze the learning time in combination with a predetermined learning memory curve to obtain the target user's memory coefficient of the arbitrary candidate mathematical knowledge point; and adjust the arbitrary relevance using the memory coefficient as a weight.

[0021] The knowledge tracking method for applied learning transfer proposed in this application first forms a historical mathematics knowledge point set based on the historical learning log of the target user, and analyzes the historical mathematics knowledge point set to obtain an initial mathematics knowledge point map. Then, the real-time mathematics knowledge point set obtained by decomposing the target course is updated to the initial mathematics knowledge point map to obtain the real-time mathematics knowledge point map. Then, the first mathematics knowledge point in the real-time mathematics knowledge point map is obtained with the real-time mathematics knowledge point set as a constraint, and the target knowledge tracking record of the target user for the first mathematics knowledge point is obtained. The target knowledge tracking record includes a first tracking record. Then, a predetermined tracking feature is read, and feature matching is performed on the first tracking record based on the predetermined tracking feature to obtain a first tracking feature parameter. Finally, a first predetermined coefficient is obtained, and the first predetermined coefficient is used as a weight to adjust and calculate the first mastery degree obtained by weighting the variation of the first tracking feature parameter to obtain a target mastery degree. The target mastery degree is used to characterize the target user's mastery of the first mathematics knowledge point, thereby achieving the technical effect of improving the accuracy of knowledge tracking and evaluation by realizing effective transfer and sharing between different courses or knowledge points. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention are briefly introduced below. Flowcharts are used in this application to illustrate the operations performed by the methods according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in precise order. Instead, the various steps may be processed in reverse order or simultaneously as needed. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0023] Figure 1 A flow chart of the knowledge tracking method for application learning transfer provided in an embodiment of the present application.

[0024] Figure 2 A schematic diagram of the process of obtaining an initial mathematical knowledge point map in the knowledge tracking method for applied learning transfer provided in an embodiment of the present application. DETAILED DESCRIPTION

[0025] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0026] In order to make the purpose, technical solutions and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limiting this application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0027] In the following description, reference is made to “some embodiments”, which describes a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict, and the terms “first\second” involved are merely used to distinguish similar objects and do not represent a specific ordering of the objects. The terms “including” and “having” and any variations are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or modules that are not clearly listed or that are inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs. The terms used herein are for the purpose of describing the embodiments of this application only.

[0028] The present application embodiment provides a knowledge tracking method for applying learning transfer, such as Figure 1 As shown, the method includes:

[0029] Step S100: Build a historical mathematical knowledge point set based on the target user's historical learning logs, and analyze the historical mathematical knowledge point set to obtain an initial mathematical knowledge point map. Specifically, data crawling technology or an API interface is used to extract the target user's (e.g., a student) historical learning logs from their learning platform or system (e.g., an online education system, a learning management system, etc.). These logs record information such as the target user's past mathematical knowledge points, exercise completion status, and study time. This information includes, but is not limited to: the user's login time and study time; the user's learning progress in different mathematics courses; the exercises and tests completed by the user and their answers (e.g., accuracy rate, answer time, error types, etc.); and the discussion forum content in which the user participated (e.g., questions asked and answers). The extracted data is preprocessed, including data cleaning (removing invalid or duplicate data) and data formatting (converting the data into a unified structured format). Based on the target user's historical learning logs, natural language processing (NLP) technology is used to analyze the text content in the learning logs (e.g., course titles, exercise descriptions, user questions, etc.) to extract all the mathematical knowledge points the user has previously learned. A predefined vocabulary of mathematical knowledge points is constructed, and knowledge points are extracted from logs through keyword matching. For example, keywords such as "linear equation" and "sum of interior angles of a triangle" are matched against the text in the logs. The extracted knowledge points are deduplicated and categorized according to the mathematical discipline classification system (such as algebra, geometry, and probability). Based on the logical relationships between knowledge points (for example, "linear equation" is a knowledge point under the "algebra" branch), a hierarchical structure of knowledge points is constructed, forming a historical mathematical knowledge point collection. This collection encompasses various mathematical concepts, formulas, theorems, and other concepts that the target user has previously encountered. This historical mathematical knowledge point collection is deeply analyzed to construct an initial mathematical knowledge point graph. Each mathematical knowledge point is defined as a node in the graph. Node attributes include the knowledge point's name, subject classification, and learning frequency. Edges between nodes are defined based on the associations between knowledge points (such as co-occurrence frequency and logical dependencies). Edge weights are calculated based on the co-occurrence frequency or cosine similarity of the knowledge point in the exercises. Use a graph database (such as Neo4j) or a graph visualization tool (such as Gephi) to store and display an initial mathematical knowledge point map. A graph is a graphical representation that intuitively shows the connections and relationships between various knowledge points. In this graph, each knowledge point is considered a node, and the lines between nodes represent the degree of connection or dependency between the knowledge points. For example, if two knowledge points frequently appear in the same problem or the same course, they will be closer in the graph, indicating a high degree of connection.

[0030] For example, user A's historical learning logs contain knowledge points such as "linear equations" and "the sum of the interior angles of a triangle." Co-occurrence frequency calculations reveal a high correlation between "linear equations" and "quadratic equations" (weighted at 0.8), while a low correlation between "linear equations" and "the sum of the interior angles of a triangle" (weighted at 0.2). Using the Gephi tool to visualize user A's mathematical knowledge map, the line connecting the "linear equations" and "quadratic equations" nodes is thicker, indicating a strong correlation between them.

[0031] like Figure 2As shown, in one possible implementation, a historical mathematical knowledge point set is constructed based on the target user's historical learning logs, and the historical mathematical knowledge point set is analyzed to obtain an initial mathematical knowledge point map. Step S100 further includes step S110, which constructs a task database containing multiple tasks tagged with mathematical knowledge points. Specifically, a large number of mathematical exercises are collected from a learning platform (via API calls), an educational resource library (using a crawler program written in a programming language such as Python to crawl exercise data), or a user's learning history (extracted from the user's learning behavior log). Each exercise is annotated with knowledge points, that is, the main mathematical knowledge points involved in each exercise are determined. The knowledge point annotating method involves constructing a predefined library containing all possible mathematical knowledge points. Using natural language processing (NLP) technology, the text content of the exercises is analyzed and automatically matched to knowledge points in the predefined knowledge point library. The automatically annotated results are manually reviewed to ensure the accuracy of the annotations. Exercises that cannot be automatically annotated are manually annotated by professional teachers. The annotated exercises and their corresponding knowledge point tags are stored in the task database, forming a set of tasks tagged with mathematical knowledge points. Step S120: Randomly obtain a first historical mathematics knowledge point and a second historical mathematics knowledge point from the set of historical mathematics knowledge points. Specifically, randomly select two mathematics knowledge points from the set of historical mathematics knowledge points as the first historical mathematics knowledge point and the second historical mathematics knowledge point for subsequent analysis. Step S130: Determine whether the first historical mathematics knowledge point and the second historical mathematics knowledge point co-occur in a first related mathematics knowledge point tag. Specifically, check whether the two historical mathematics knowledge points appear simultaneously in the related mathematics knowledge point tag of the same task (i.e., exercise). If so, it indicates that the two knowledge points are related to some extent. Step S140: If they co-occur, reversely match the first task corresponding to the first related mathematics knowledge point tag and add the first task to the co-occurrence task list. Specifically, if the two historical mathematics knowledge points co-occur, reversely find an exercise containing the two knowledge points and use it as a co-occurrence task. The found co-occurrence task is added to the co-occurrence task list, which is used to record which tasks simultaneously involve the two knowledge points. Step S150, using the ratio of the number of tasks in the co-occurring task list to the number of tasks in the task database as the correlation between the first historical mathematics knowledge point and the second historical mathematics knowledge point. Specifically, the correlation between the two historical mathematics knowledge points is evaluated by calculating the ratio of the number of tasks in the co-occurring task list to the total number of tasks in the task database. The higher the ratio, the higher the frequency of the two knowledge points in the tasks, and therefore the higher the correlation between them. Step S160, using the correlation as a constraint, combining the first historical mathematics knowledge point and the second historical mathematics knowledge point to construct the initial mathematics knowledge point map.Specifically, using the calculated correlation as a constraint, an initial mathematical knowledge point map is constructed by combining various historical mathematical knowledge points. In this map, the distance between knowledge points reflects their correlation. Knowledge points with higher co-occurrence frequencies (i.e., greater correlation) are closer in the map. This implementation method infers the degree of association between knowledge points by analyzing their co-occurrence in tasks, thereby constructing a map that reflects the intrinsic connections between knowledge points, providing a foundation for subsequent learning transfer and knowledge tracking.

[0032] In one possible implementation, the ratio of the number of tasks in the co-occurring task list to the number of tasks in the task database is used as the correlation between the first historical mathematics knowledge point and the second historical mathematics knowledge point. Step S150 further includes step S151: obtaining a first history vector for the first historical mathematics knowledge point. Specifically, all learning tasks (i.e., exercises) related to the first historical mathematics knowledge point and their completion status (e.g., accuracy rate, time required, etc.) are extracted from historical learning logs. Based on these learning tasks and their completion status, features related to the first historical mathematics knowledge point are extracted, such as accuracy rate, error types, and learning time. The extracted features are quantized and combined into a multidimensional vector, which serves as the history vector for the first historical mathematics knowledge point. Step S152: obtaining a second history vector for the second historical mathematics knowledge point. Similarly, all learning tasks related to the second historical mathematics knowledge point and their completion status are extracted from historical learning logs. Based on these learning tasks and their completion status, features related to the second historical mathematics knowledge point are extracted. The extracted features are quantized and combined into a multidimensional vector, which serves as the history vector for the second historical mathematics knowledge point. In step S153, the similarity between the first historical vector and the second historical vector is calculated using the cosine similarity principle. Specifically, the two historical vectors are normalized so that their lengths (i.e., the vector moduli) are 1. Then, the dot product of the two normalized historical vectors is calculated and divided by the product of their moduli to obtain the cosine similarity. Cosine similarity measures the degree of directional similarity between two vectors and has a value range of [-1, 1]. When the two vectors have exactly the same directions, the cosine similarity is 1; when the two vectors have completely opposite directions, the cosine similarity is -1; and when the two vectors are perpendicular, the cosine similarity is 0. In step S154, the correlation is adjusted using the similarity as a weight. Specifically, a weight value is determined based on the cosine similarity obtained in step S153. This weight value can be obtained by performing a linear or nonlinear transformation based on the specific value of the cosine similarity. The co-occurrence frequency-based correlation calculated in step S150 is multiplied by the weight value obtained in step S153 to obtain the adjusted correlation. The adjusted correlation not only considers the co-occurrence frequency between knowledge points, but also the similarity of their historical performance during the learning process. This implementation captures the similarity of the historical performance of knowledge points during the learning process by constructing history vectors and calculating the cosine similarity between them. It comprehensively assesses the degree of association between historical mathematical knowledge points. When two historical mathematical knowledge points have a high co-occurrence frequency and similar historical performance during the learning process, the degree of association between the two knowledge points is considered to be strong. At the same time, by introducing cosine similarity as a weight to adjust the correlation, it reduces to a certain extent the deviation in the co-occurrence frequency calculation caused by factors such as data sparsity and noise, thereby improving the accuracy and robustness of the correlation assessment.

[0033] Step S200 updates the real-time mathematical knowledge point set obtained by decomposing the target course into the initial mathematical knowledge point map, thereby obtaining a real-time mathematical knowledge point map. Specifically, when a target user begins learning a new course or knowledge point, the target course (e.g., the currently studied mathematics course) is decomposed into multiple mathematical knowledge points to form a real-time mathematical knowledge point set. The knowledge points in the real-time mathematical knowledge point set are added to the initial mathematical knowledge point map, and relevance information (e.g., the relationship between the newly added knowledge point and other knowledge points) is updated to obtain a real-time mathematical knowledge point map. This map is continuously updated and expanded as the target user progresses, reflecting the target user's current knowledge mastery in real time. For example, if a student begins learning "inequalities" after learning "linear equations" and "quadratic equations," a new node representing "inequalities" will be added to the updated real-time mathematical knowledge point map, and corresponding connections will be added to the map based on the degree of relevance between "inequalities" and other knowledge points (e.g., "linear equations").

[0034] Step S300: Using the real-time mathematical knowledge point set as a constraint, obtain a first mathematical knowledge point in the real-time mathematical knowledge point graph, and obtain the target knowledge tracking record of the target user for the first mathematical knowledge point, wherein the target knowledge tracking record includes the first tracking record. Specifically, in the real-time mathematical knowledge point graph, select one as the first mathematical knowledge point (i.e., a new knowledge point or a knowledge point currently being learned) based on the current real-time mathematical knowledge point set (i.e., the knowledge point the user is currently learning). Find exercises (tasks) related to the first mathematical knowledge point, monitor the target user's progress in completing these exercises, and record the completion status (e.g., accuracy, time consumption, etc.) to form a target knowledge tracking record. In other words, the target knowledge tracking record reflects the target user's completion of related exercises while learning the first mathematical knowledge point.

[0035] In one possible implementation, a first mathematical knowledge point in the real-time mathematical knowledge point graph is obtained using the real-time mathematical knowledge point set as a constraint, and the target user's target knowledge tracking record for the first mathematical knowledge point is obtained. Step S300 further includes step S310: extracting a second task from the multiple tasks with mathematical knowledge point tags, where the second task corresponds to the second mathematical knowledge point tag. Specifically, a task database storing mathematical knowledge point tags is accessed and all tasks containing mathematical knowledge point tags are filtered from the task database. One of these tasks is selected as the second task based on a certain strategy (such as random selection), and the second task is associated with one or more mathematical knowledge point tags. Step S320: Determine whether the first mathematical knowledge point belongs to the second mathematical knowledge point tag. Specifically, the mathematical knowledge point tags associated with the second task (i.e., the second mathematical knowledge point tags) are extracted from the second task, and the first mathematical knowledge point is matched with the second mathematical knowledge point tags to determine whether the first mathematical knowledge point is included in these tags. Step S330: If so, the second task is added to a candidate task list. Specifically, if the first mathematical knowledge point belongs to the second mathematical knowledge point label, that is, the first mathematical knowledge point is related to the second task, the second task is added to the candidate task list, which is used to store tasks related to the first mathematical knowledge point. Step S340, monitors the process of the target user processing each task in the candidate task list to obtain the target knowledge tracking record. Specifically, the tasks in the candidate task list are assigned to the target user for practice or answering. Using technical means such as a learning management system or an intelligent teaching system, the process of the target user processing these tasks is monitored in real time, and various data of the target user in the process of processing the task are recorded, such as accuracy, answering time, error type, etc. These data constitute the target knowledge tracking record. This implementation method accurately obtains the target user's learning and mastery of the knowledge point by monitoring the target user's process of processing tasks related to the current learning knowledge point.

[0036] In one possible implementation, the target user's progress in handling each task in the candidate task list is monitored to obtain the target knowledge tracking record. Step S340 further includes step S341: analyzing the real-time mathematical knowledge point graph with the first mathematical knowledge point as the center and a predetermined distance as the radius to determine a first associated mathematical knowledge point range for the first mathematical knowledge point. Specifically, the first mathematical knowledge point currently being learned is used as the center point for analysis, and a radius is set based on a predetermined distance (this distance can be a fixed value or dynamically adjusted based on the complexity of learning transfer and the relevance of the knowledge points). In the real-time mathematical knowledge point graph, the graph is searched for other mathematical knowledge points associated with the first mathematical knowledge point, using the center point and radius as constraints. All searched associated mathematical knowledge points form a range, namely, the first associated mathematical knowledge point range for the first mathematical knowledge point. Step S342: extracting any associated mathematical knowledge points within the first associated mathematical knowledge point range. Specifically, all associated mathematical knowledge points within the first associated mathematical knowledge point range are traversed. A related mathematical knowledge point is randomly extracted from the range or according to a certain strategy (e.g., based on the strength of the association) as the current processing target. Step S343 determines whether the arbitrary related mathematical knowledge point belongs to the third related mathematical knowledge point tag. Specifically, the related mathematical knowledge point tag of the currently processed related mathematical knowledge point is obtained, and it is determined whether the tag matches a predetermined third related mathematical knowledge point tag. Step S344: If so, a third task corresponding to the third related mathematical knowledge point tag is added to the candidate task list. Specifically, if the tag of the currently processed related mathematical knowledge point matches the third related mathematical knowledge point tag, a task associated with the tag is searched and added to the candidate task list for subsequent monitoring and evaluation of the target user. This implementation method increases the relevance of learning transfer by searching for other mathematical knowledge points associated with the first mathematical knowledge point and adding the tasks corresponding to these knowledge points to the candidate task list, making the learning materials more relevant to the target user's current learning needs. By monitoring the target user's progress in processing these related tasks, more data can be collected regarding the target user's mastery of the first mathematical knowledge point and its related knowledge points, thereby improving the accuracy of the evaluation of the learning transfer effect.

[0037] Step S400 reads predetermined tracking features and performs feature matching on the first tracking record based on the predetermined tracking features to obtain first tracking feature parameters. Specifically, certain tracking features are predefined, including question accuracy, which reflect the user's mastery of the knowledge point. These predetermined tracking features are read and feature matched with the first tracking record in the target knowledge tracking record to obtain first tracking feature parameters, such as accuracy parameters and time parameters.

[0038] In one possible implementation, step S400 further includes step S410, and the predetermined tracking features include task accuracy and unit task time. Specifically, the task accuracy is calculated by counting the target user's answers to the first mathematical knowledge point in exercises, tests or homework. For example, if the user correctly answers 8 times in 10 exercises, then the task accuracy is 80%. Unit task time refers to the average time, that is, the sum of the time of all related tasks divided by the number of tasks. For example, if the user spends a total of 100 minutes in 10 exercises, then the average time per exercise is 10 minutes. This implementation method comprehensively evaluates the target user's mastery of the first mathematical knowledge point by combining task accuracy and unit task time. Relying solely on accuracy will ignore the differences in users' time management or learning efficiency. Adding unit task time can more carefully reflect the user's learning status.

[0039] Step S500: Obtain a first predetermined coefficient and, using the first predetermined coefficient as a weight, adjust the first mastery level obtained by weighting the first tracking feature parameter using the variational weighting to obtain a target mastery level. The target mastery level represents the target user's mastery of the first mathematical knowledge point. Specifically, the first predetermined coefficient is obtained. The first predetermined coefficient is a weighting coefficient derived from the task transfer index and used to adjust the calculation of the target mastery level. Since the target knowledge tracking record includes more than one tracking record, a variational weighting process is performed on all first tracking feature parameters, including standardization and weighted summation of each feature parameter. This yields a comprehensive, overall performance indicator, namely, the target user's initial mastery level of the first mathematical knowledge point, also known as the first mastery level. The first mastery level is a quantitative value that reflects the target user's overall performance in learning the first mathematical knowledge point. The following is an example of calculating the first mastery level: Each first tracking feature parameter is normalized to a uniform range of [0, 1]. The first tracking feature parameters include accuracy, completion time, and number of attempts, with weights of 0.5, 0.3, and 0.2, respectively. Accuracy = 0.8, Completion Time = 120 seconds, Number of Attempts = 3. The standardized values are: Accuracy = 0.8, Completion Time = 0.6 (maximum time is 200 seconds), Number of Attempts = 0.5 (maximum number of attempts is 6). First Mastery = 0.5 * 0.8 + 0.3 * 0.6 + 0.2 * 0.5 = 0.4 + 0.18 + 0.1 = 0.68. After obtaining the first mastery, adjust it based on the first predetermined coefficient. Multiply the first mastery by the first predetermined coefficient to obtain the final target mastery. This metric represents the target user's mastery of the first mathematical knowledge point and can serve as a basis for subsequent learning recommendations, evaluation, or feedback. The embodiment of the present application adopts the following technical means: forming a historical mathematics knowledge point set based on the historical learning log of the target user, and analyzing it to obtain an initial mathematics knowledge point map, updating the real-time mathematics knowledge point set obtained by decomposing the target course to the initial mathematics knowledge point map, obtaining the real-time mathematics knowledge point map, obtaining the first mathematics knowledge point in the real-time mathematics knowledge point map with the real-time mathematics knowledge point set as a constraint, obtaining the target knowledge tracking record of the target user for the first mathematics knowledge point, reading the predetermined tracking feature, and performing feature matching on the first tracking record based on the predetermined tracking feature to obtain a first tracking feature parameter, obtaining a first predetermined coefficient, and using the first predetermined coefficient as a weight to adjust and calculate the first mastery degree obtained by weighting the variation of the first tracking feature parameter to obtain the target mastery degree, etc., thereby achieving the technical effect of improving the accuracy of knowledge tracking and evaluation by realizing effective migration and sharing between different courses or knowledge points.

[0040] In one possible implementation, a first predetermined coefficient is obtained, and the first mastery level obtained by weighting the variation of the first tracking feature parameter is adjusted and calculated using the first predetermined coefficient as a weight to obtain a target mastery level. Step S500 further includes step S510: reverse matching a first candidate task corresponding to the first tracking record, wherein the first candidate task corresponds to a first candidate mathematical knowledge point tag. Specifically, learning logs, task records, or user interaction history with the system are reviewed, and task identifiers or knowledge point tags are used for reverse matching to identify learning tasks directly related to the first tracking record, i.e., the first candidate task. Step S520: randomly extracting any candidate mathematical knowledge point from the first candidate mathematical knowledge point tag. Specifically, after determining the first candidate task, all mathematical knowledge point tags involved in the first candidate task are extracted. A random selection algorithm is used to randomly select one of these knowledge point tags as a candidate mathematical knowledge point. The candidate mathematical knowledge point is used to evaluate the relevance and transfer potential of the first mathematical knowledge point with other knowledge points. Step S530: combining the real-time mathematical knowledge point graph to obtain any correlation between the candidate mathematical knowledge point and the first mathematical knowledge point. Specifically, the real-time mathematical knowledge point graph is used to calculate the correlation between any candidate mathematical knowledge point and the first mathematical knowledge point. The correlation is calculated based on the distance between the knowledge points in the graph. The closer the distance, the higher the correlation. The correlation can be calculated using the following formula: Correlation = ; Wherein, distance refers to the shortest path length between two knowledge points in the mathematical knowledge point map. Step S540, take the mean of the arbitrary correlation as the first task migration index of the first candidate task, and record the first task migration index as the first predetermined coefficient. Specifically, calculate the mean of the correlation between all randomly extracted arbitrary candidate mathematical knowledge points and the first mathematical knowledge point. Use this mean as the first task migration index of the first candidate task, which reflects the difficulty of migrating from other related knowledge points to the first mathematical knowledge point. Record the first task migration index as the first predetermined coefficient, which is used for the adjustment calculation of the first mastery. This implementation method accurately evaluates the difficulty that the user may encounter when learning new knowledge points by calculating the task migration index (i.e., the first predetermined coefficient), improves the accuracy of the adjustment of the first mastery, and thereby improves the accuracy of the target mastery.

[0041] The following is a specific example of obtaining the first predetermined coefficient:

[0042] Table 1 shows the learning record of a target user for "linear equations of one variable":

[0043] Table 1 Learning records

[0044] Task ID Knowledge points involved Completion time T001 Linear equation of one variable 2024-01-01 T002 Quadratic equations 2024-01-05 T003 First degree equation of one variable, sum of interior angles of triangle 2024-01-10

[0045] The target knowledge point currently being evaluated, the first mathematical knowledge point, is "Linear Equations." Table 1 is traversed to find tasks containing "Linear Equations." Matched tasks are T001 and T003. For task T001, the randomly selected knowledge point is "Linear Equations." For task T003, the randomly selected knowledge point is "Sum of the Interior Angles of a Triangle." Table 2 shows the real-time mathematical knowledge point map:

[0046] Table 2 Real-time mathematical knowledge point map

[0047] Knowledge Point A Knowledge Point B Shortest path length Linear equation of one variable Linear equation of one variable 0 Linear equation of one variable sum of interior angles of triangle 3

[0048] For Task T001: Knowledge point A = linear equation, Knowledge point B = linear equation, Shortest path length = 0, Correlation = =1.0. For Task T003: Knowledge Point A = Linear Equation, Knowledge Point B = Sum of Triangle Interior Angles, Shortest Path Length = 3, Relevance = =0.25. Calculate the mean of all correlations: First predetermined coefficient = First task transfer index = =0.625.

[0049] In one possible implementation, after obtaining any correlation between the candidate mathematical knowledge point and the first mathematical knowledge point in conjunction with the real-time mathematical knowledge point graph, step S530 further includes step S531: obtaining the learning time of the candidate mathematical knowledge point. Specifically, the target user's learning history or relevant data in a learning management system (LMS) is accessed to obtain the target user's learning time for each candidate mathematical knowledge point. Specifically, the completion time of each candidate mathematical knowledge point is extracted from the target user's learning history, the current time is recorded using system time or server time, and the current time is subtracted from the completion time of each candidate mathematical knowledge point to obtain the learning time. The learning time refers to the length of time from the target user's completion of learning a particular mathematical knowledge point to the current time. All randomly extracted candidate mathematical knowledge points are traversed, and corresponding learning time data is collected for each knowledge point. Step S532: The learning time is analyzed in conjunction with a predetermined learning and memory curve to obtain the target user's memory coefficient for the candidate mathematical knowledge point. Specifically, a predetermined learning and memory curve (such as the Ebbinghaus forgetting curve) is used to analyze the learning time of each candidate mathematical knowledge point, and a memory coefficient is calculated accordingly. The learning and memory curve describes the law of human memory decay over time. Through this curve, the user's memory retention of the knowledge point after a period of time can be estimated. The memory coefficient is a value between 0 and 1, indicating the user's memory retention ratio of the knowledge point. For example, the Ebbinghaus forgetting curve is used as the mathematical model of the learning and memory curve. The curve can be expressed by the following formula: ;in, represents the memory retention ratio at time t, and τ is a constant representing the time constant of memory decay. Based on the correspondence between the learning time and the learning-memory curve, a memory coefficient is calculated for each candidate mathematical knowledge point. Specifically, the learning time is substituted into the learning-memory curve formula to calculate the memory retention ratio. The memory retention ratio is the memory coefficient, which represents the target user's memory retention of any candidate mathematical knowledge point. The longer the learning time, the lower the memory coefficient, indicating that the user's memory retention of the knowledge point is lower.

[0050] Step S533, adjusting the arbitrary relevance using the memory coefficient as a weight. Specifically, after obtaining the memory coefficient of each arbitrary candidate mathematical knowledge point, these coefficients are used as weights to adjust the arbitrary relevance calculated previously. The adjusted relevance better reflects the actual difficulty of the target user migrating from any candidate mathematical knowledge point to the first mathematical knowledge point at the current time point. Knowledge points with higher memory coefficients (i.e., knowledge points that are better retained by users) are easier to migrate, so their relevance will be improved accordingly; conversely, knowledge points with lower memory coefficients are more difficult to migrate, so their relevance will be reduced accordingly. Finally, the first task migration index of the first candidate task is calculated using the adjusted relevance. This implementation method accurately evaluates the actual difficulty of migrating from any candidate mathematical knowledge point to the first mathematical knowledge point through the learning time and memory coefficient of any candidate mathematical knowledge point, thereby improving the accuracy of obtaining the mastery of the first mathematical knowledge point.

[0051] The above specific embodiments do not constitute a limitation to the scope of protection of this application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements and improvements made within the spirit and principles of this application should be included in the scope of protection of this application. In some cases, the actions or steps recorded in this application can be performed in an order different from that in the embodiments and can still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. The knowledge tracking method of applied learning transfer is characterized by: include: Build a historical mathematics knowledge point set based on the historical learning log of the target user, and analyze the historical mathematics knowledge point set to obtain an initial mathematics knowledge point map; Updating the real-time mathematics knowledge point set obtained by decomposing the target course into the initial mathematics knowledge point map to obtain a real-time mathematics knowledge point map; Acquire a first mathematical knowledge point in the real-time mathematical knowledge point graph using the real-time mathematical knowledge point set as a constraint, and acquire a target knowledge tracking record of the target user for the first mathematical knowledge point, wherein the target knowledge tracking record includes a first tracking record; Reading a predetermined tracking feature, and performing feature matching on the first tracking record based on the predetermined tracking feature to obtain a first tracking feature parameter; Obtaining a first predetermined coefficient, and using the first predetermined coefficient as a weight, adjusting and calculating a first mastery degree obtained by weighting the variation of the first tracking feature parameter to obtain a target mastery degree, wherein the target mastery degree is used to represent the target user's mastery degree of the first mathematical knowledge point, and the first mastery degree is the target user's initial mastery degree of the first mathematical knowledge point. The first predetermined coefficient is a weight coefficient obtained based on the task transfer index and is used to adjust the calculation of the target mastery degree, including: Reverse matching a first candidate task corresponding to the first tracking record, wherein the first candidate task corresponds to a first candidate mathematical knowledge point label; Randomly extract any candidate mathematical knowledge point from the first candidate mathematical knowledge point label; Combining the real-time mathematical knowledge point map to obtain any correlation between the arbitrary candidate mathematical knowledge point and the first mathematical knowledge point; An average of the arbitrary correlations is taken as a first task transition index of the first candidate task, and the first task transition index is recorded as the first predetermined coefficient.

2. The knowledge tracking method for application learning transfer according to claim 1 is characterized in that: A historical mathematics knowledge point set is formed based on the historical learning log of the target user, and the historical mathematics knowledge point set is analyzed to obtain an initial mathematics knowledge point map, including: Establishing a task database, wherein the task database includes a plurality of tasks having labels related to mathematical knowledge points; Randomly obtaining a first historical mathematics knowledge point and a second historical mathematics knowledge point from the set of historical mathematics knowledge points; Determining whether the first historical mathematics knowledge point and the second historical mathematics knowledge point co-appear in a first related mathematics knowledge point label; If it is co-occurrence, reversely match the first task corresponding to the first mathematical knowledge point label, and add the first task to the co-occurrence task list; using the ratio of the number of tasks in the co-occurring task list to the number of tasks in the task database as the relevance between the first historical mathematics knowledge point and the second historical mathematics knowledge point; Taking the relevance as a constraint, the initial mathematics knowledge point map is constructed by combining the first historical mathematics knowledge point and the second historical mathematics knowledge point.

3. The knowledge tracking method for application learning transfer according to claim 2 is characterized in that: The method further includes: using the ratio of the number of tasks in the co-occurring task list to the number of tasks in the task database as the relevance between the first historical mathematics knowledge point and the second historical mathematics knowledge point; Obtaining a first history vector of the first historical mathematical knowledge point; Obtaining a second history vector of the second historical mathematical knowledge point; Obtaining the similarity between the first history vector and the second history vector using the cosine similarity principle; The relevance is adjusted using the similarity as a weight.

4. The knowledge tracking method for application learning transfer according to claim 2 is characterized in that: Acquiring a first mathematical knowledge point in the real-time mathematical knowledge point graph using the real-time mathematical knowledge point set as a constraint, and acquiring a target knowledge tracking record of the target user for the first mathematical knowledge point, including: Extracting a second task from the plurality of tasks having the mathematics knowledge point related labels, wherein the second task corresponds to a second mathematics knowledge point related label; Determining whether the first mathematical knowledge point belongs to the second mathematical knowledge point label; If yes, add the second task to the candidate task list; The process of the target user processing each task in the candidate task list is monitored to obtain the target knowledge tracking record.

5. The knowledge tracking method for application learning transfer according to claim 4 is characterized in that: Monitoring the process of the target user processing each task in the candidate task list to obtain the target knowledge tracking record further includes: Analyzing the real-time mathematical knowledge point map with the first mathematical knowledge point as the center and a predetermined distance as the radius to determine a range of first mathematical knowledge points associated with the first mathematical knowledge point; Extracting any associated mathematical knowledge point within the first associated mathematical knowledge point range; Determining whether the any associated mathematical knowledge point belongs to a third related mathematical knowledge point label; If yes, add the third task corresponding to the third mathematical knowledge point label to the candidate task list.

6. The knowledge tracking method for application learning transfer according to claim 1, characterized in that: The predetermined tracking features include task accuracy and unit task time.

7. The knowledge tracking method for application learning transfer according to claim 1, characterized in that: Combining the real-time mathematical knowledge point map to obtain any correlation between the arbitrary candidate mathematical knowledge point and the first mathematical knowledge point, and then further comprising: Obtaining the learning time of any candidate mathematical knowledge point; Analyzing the learning time in combination with a predetermined learning and memory curve to obtain a memory coefficient of the target user for the arbitrary candidate mathematical knowledge point; The arbitrary correlation is adjusted using the memory coefficient as a weight.

Images