A test and evaluation method for the English learning process

Abstract

The present invention discloses a test evaluation method for an English learning process, and relates to the field of adaptive learning systems. The present invention comprises the following steps: S1. constructing a hierarchical knowledge graph based on knowledge point dependency relationships; when a student completes a task or a test, updating the knowledge point mastery in the knowledge graph in real time; S2. inputting the updated knowledge graph into a DTN network model, and the DTN network model automatically determines the tasks that the student currently needs to learn by combining direct dependency relationships with indirect dependency relationships; S3. by calculating the student's knowledge mastery and path dependency matrix, recommending the most suitable learning content and exercises for the student's current level. The present invention realizes indirect dependency modeling between knowledge points by introducing a path dependency matrix and a deep temporal network model, and improves the intelligence and personalization level of the adaptive language learning system.

Description

Technical Field

[0001] The present invention relates to the field of adaptive learning systems, and specifically to a test and evaluation method for the English learning process. Background Art

[0002] With the rapid development of educational technology, especially the breakthroughs in the fields of artificial intelligence and deep learning, many adaptive learning systems have entered practical applications, especially in the field of language learning. The computerized adaptive test (CAT) based on item response theory has gradually replaced the classical test and is becoming more and more popular at home and abroad. CAT is a test method that adapts to the ability level of candidates. The computer-aided examination tailors the test questions according to the ability of the candidates. In the field of computerized adaptive test (CAT), item response theory is usually used to select the most suitable items according to the individual ability of the candidates. In a computerized adaptive test (CAT), each candidate takes a unique test tailored to their ability level. The selection of computer-aided translation subjects is a selection based on the individual ability level of the candidates, rather than a single fixed examination. After each response, the ability estimate is updated, and the next item is selected to have optimal properties according to the new estimate.

[0003] Although many achievements have been made in the existing adaptive learning systems, there are still some limitations, which are mainly reflected in the following aspects. First of all, the existing technologies are mostly limited to personalized recommendation and learning path adjustment based on direct dependencies, lacking effective modeling of the indirect dependencies between knowledge points. Most systems recommend learning content based on simple test results and students' mastery, but rarely consider the complex dependencies between knowledge points, especially indirect dependencies. In the actual learning process, the mastery of some knowledge points is not directly judged by test scores, but depends on the mastery of other related knowledge. For example, if a student does not master the "grammar" knowledge point well, it may affect their performance in "reading comprehension". The neglect of this kind of indirect dependency in the existing technology leads to the recommended learning content not fully meeting the actual needs of students, thus affecting the learning effect. Summary of the Invention

[0004] The present invention proposes a test and evaluation method for the English learning process. Based on the dynamic learning path adjustment ability of the deep learning model, it can respond to the changes in the learning state of students in real time, flexibly adjust the learning path, and solve the inefficiency problem caused by the static learning path in the existing technology.

[0005] Among them, a test and evaluation method for the English learning process includes the following steps:

[0006] S1. Construct a hierarchical knowledge graph based on knowledge point dependencies. The hierarchical knowledge graph organizes the core knowledge points of language learning by levels and dynamically updates the mastery level of each knowledge point. When a student completes a task or a test, the mastery level of the knowledge points in the knowledge graph is updated in real time;

[0007] S2. Input the newly updated knowledge graph into the DTN network model. The DTN network model combines direct dependencies and indirect dependencies to identify potential knowledge points that the student has not mastered and automatically determines the tasks that the student currently needs to learn;

[0008] S3. According to the intelligent diagnosis results, by calculating the student's knowledge mastery level and the path dependency matrix, recommend the learning content and exercises that are most suitable for the student's current level, and optimize the learning path;

[0009] Among them, the specific steps of step S1 include the following sub-steps:

[0010] S101. Initialize the construction of the knowledge graph, representing the knowledge graph as a directed graph composed of a node set and an edge set. The nodes represent the core knowledge points, and the edges represent the dependencies between the knowledge points;

[0011] S102. According to the initialized knowledge graph information, represent the direct dependencies between the core knowledge points by constructing a path dependency matrix;

[0012] S103. Obtain the indirect dependencies through the power operation of the matrix;

[0013] S104. Update the mastery level of the knowledge points according to the student's performance in the test and learning tasks.

[0014] Furthermore, the core knowledge points include basic information, prerequisite knowledge point information, and post-requisite knowledge point information; the basic information is the basic attributes, including the unique identifier, name, and the hierarchical position of the knowledge point in the knowledge system; the prerequisite knowledge point information is the knowledge points that need to be mastered before learning the current knowledge point; the post-requisite knowledge point information is other knowledge points that depend on the current knowledge point.

[0015] Furthermore, the path dependency matrix is a matrix, where each element represents whether the core knowledge point is a prerequisite knowledge point of the core knowledge point ; when the knowledge point is a prerequisite knowledge point of the knowledge point , , otherwise .

[0016] Further, in the step S103, the specific calculation process of the indirect dependence relationship is represented by the transmission of the step dependence relationship:

[0017] ;

[0018] wherein, the represents the index of the power, and the represents the path dependence matrix.

[0019] Further, the step S2 specifically includes the following sub-steps:

[0020] S201. Use the core knowledge points as task nodes, the dependence relationship between knowledge points as edges, and the dependence strength calculated according to the mastery degree and indirect dependence relationship as weights to establish a DTN network model;

[0021] S202. The DTN network model performs direct dependence analysis according to the direct dependence relationship and indirect dependence analysis according to the indirect dependence relationship respectively;

[0022] S203. Push learning tasks according to the analysis results in combination with the indirect dependence relationship.

[0023] Further, in the step S202, the specific calculation process of the direct dependence analysis is to evaluate the influence of the direct dependence for each knowledge point according to the path dependence matrix and calculate the mastery situation of the trainee for the directly dependent knowledge points, that is:

[0024] ;

[0025] wherein, the represents the weighted sum of the mastery degrees of the knowledge points directly dependent on the knowledge point, the represents the index of the target knowledge point, the represents the direct dependence relationship between the starting knowledge point and the target knowledge point, and the represents the mastery degree of the target knowledge point.

[0026] Further, in the step S202, the specific calculation process of the indirect dependence analysis is to calculate the mastery degree of each knowledge point through the indirect path dependence and calculate the potential indirect knowledge points mastered by the trainee based on the indirect dependence relationship:

[0027] ;

[0028] wherein, the represents the mastery degree of the potential knowledge points through k-step dependence, the represents the index of the power, and the represents the indirect dependence relationship between the starting knowledge point and the target knowledge point, Indicates the mastery level of the target knowledge point.

[0029] Furthermore, step S3 specifically includes the following sub-steps:

[0030] S301. Calculate the learning path that best suits the current learning level of the student based on the student's current knowledge point mastery level and the path dependence matrix;

[0031] S302. Recommend corresponding learning content and exercises according to the learning path;

[0032] S303. Dynamically adjust the learning path according to the student's learning progress.

[0033] Furthermore, in step S301, the specific calculation process for calculating the learning path that best suits the current learning level of the student is as follows: for each knowledge point, calculate the path from the knowledge point reached by the current learning level to the target knowledge point, that is:

[0034] ;

[0035] Among them, the represents the learning path score from the starting knowledge point to the target knowledge point, the represents the mastery level of the knowledge point reached by the current learning level, the represents the dependence relationship from the starting knowledge point to the knowledge point reached by the current learning level, and the represents the dependence relationship from the knowledge point reached by the current learning level to the target knowledge point.

[0036] Furthermore, it also includes step S4: After each learning cycle, train and update the DTN network model through the student's historical learning data, optimize the path dependence matrix according to the student's learning progress and feedback, and re-evaluate the student's learning status by combining the output of the DTN network model and the dynamically adjusted path dependence matrix, and adjust the learning path and content recommendation.

[0037] The beneficial effects of the invention are:

[0038] By introducing the path dependence matrix and the deep temporal network (DTN) model, the present invention realizes the indirect dependence modeling between knowledge points, greatly improving the intelligence and personalization level of the adaptive learning system. By closely combining the path dependence relationship with the student's learning dynamics, the system can not only more comprehensively evaluate the student's knowledge mastery, but also accurately predict the possible future learning difficulties of the student. This method can ensure that the learning content recommendation is more in line with the actual learning needs of the student, thereby improving the learning efficiency and effect. Brief Description of the Drawings

[0039] Figure 1 This is a flowchart of a test and evaluation method for the English learning process provided by an embodiment of the present invention. Specific embodiments

[0040] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings, but the protection scope of the present invention is not limited to the following description.

[0041] In order to make the purpose, technical solution and advantages of the present invention clearer, the present invention will be further described in detail in combination with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention, that is, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Usually, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0042] Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed present invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present invention. It should be noted that relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations.

[0043] Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or mechanical device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or mechanical device. Without further limitation, an element defined by the statement "including a..." does not exclude the existence of additional identical elements in the process, method, article or mechanical device including the element.

[0044] The features and performance of the present invention will be further described in detail below in combination with embodiments.

[0045] Among them, as Figure 1 , a test and evaluation method for the English learning process includes the following steps:

[0046] S1. Construct a hierarchical knowledge graph based on the dependency relationship of knowledge points. The hierarchical knowledge graph organizes the core knowledge points of language learning by level and dynamically updates the mastery degree of each knowledge point; when the trainee completes the task or test, the mastery degree of the knowledge points in the knowledge graph is updated in real time;

[0047] S2. Input the newly updated knowledge graph into the DTN network model. The DTN network model combines direct and indirect dependencies to identify potential knowledge points that the learner has not yet mastered and automatically determines the tasks that the learner currently needs to learn.

[0048] S3. Based on the intelligent diagnosis results, by calculating the learner's knowledge mastery degree and path dependency matrix, recommend the most suitable learning content and exercises for the learner's current level, and optimize the learning path.

[0049] Among them, step S1 specifically includes the following sub-steps:

[0050] S101. Initialize and construct the knowledge graph, representing the knowledge graph as a directed graph composed of a node set and an edge set. Nodes represent core knowledge points, and edges represent the dependency relationships between knowledge points.

[0051] S102. According to the initialized knowledge graph information, represent the direct dependency relationships between core knowledge points by constructing a path dependency matrix.

[0052] S103. Obtain the indirect dependency relationships through matrix power operations.

[0053] S104. Update the mastery degree of knowledge points according to the learner's performance in tests and learning tasks.

[0054] Furthermore, the core knowledge points include basic information, prerequisite knowledge point information, and post-requisite knowledge point information. The basic information is basic attributes, including a unique identifier, name, and the hierarchical position of this knowledge point in the knowledge system. The prerequisite knowledge point information is the knowledge points that need to be mastered before learning the current knowledge point. The post-requisite knowledge point information is other knowledge points that depend on the current knowledge point.

[0055] Specifically, in step S1 of the above embodiments, the mastery degree refers to the mastery situation of a certain knowledge point by the trainee, which is measured by the test score or the completion situation of the learning task. The mastery degree is a specific value (such as a score between 0 and 1), indicating the mastery degree of the trainee for a certain knowledge point. In step S2, the DTN (Deep Temporal Network) model is used to predict the knowledge points that the trainee may master in the future learning process. The DTN network model analyzes the evolution of the learning state based on the trainee's historical learning data (such as previous mastery degrees, task performances, learning progress, etc.), and predicts which knowledge points the trainee will master next. The mastery degree in step S1 is a static and immediate indicator, reflecting the mastery situation of the trainee for a specific knowledge point, and is calculated and updated based on the trainee's actual performance (such as test results, task completion situations, etc.). The prediction of the DTN network model is a dynamic prediction model, which predicts which knowledge points the trainee will master in the future learning process according to the trainee's historical learning data and path dependence relationship, and determines the trainee's learning tasks according to the prediction results.

[0056] Further, the path dependence matrix is a matrix, where each element represents whether the core knowledge point is a prerequisite knowledge point for the core knowledge point ; when the knowledge point is a prerequisite knowledge point for the knowledge point , , otherwise .

[0057] Exemplarily, in English learning, a certain "grammar" knowledge point depends on the mastery of "vocabulary", and "reading comprehension" depends on "grammar". In addition, the specific structure of the path dependence matrix is:

[0058] , where the A represents the path dependence matrix.

[0059] Exemplarily, there are the following knowledge points: vocabulary (v1), grammar (v2), reading comprehension (v3). Among them, vocabulary is a prerequisite knowledge point for grammar and reading comprehension; grammar is a prerequisite knowledge point for reading comprehension. Then the path dependence matrix can be expressed as:

[0060] ;

[0061] This matrix represents:

[0062] Vocabulary (v1) is a prerequisite for grammar (v2) and reading comprehension (v3);

[0063] Grammar (v2) is a prerequisite for reading comprehension (v3);

[0064] Reading Comprehension (v3) has any prerequisite dependencies.

[0065] Furthermore, in step S103, the specific calculation process of the indirect dependency relationship is represented by the transmission of step dependency relationships:

[0066] ;

[0067] Among them, the represents the index of the power, and the represents the path dependency matrix. Exemplarily, based on the initial matrix (i.e., the path dependency matrix), the indirect dependency relationship can be obtained through matrix power operation. The element of the matrix represents the indirect dependency relationship of knowledge point to (i.e., indirectly affected through other knowledge points). Therefore, it is necessary to calculate the second power of the path dependency matrix, that is:

[0068] ;

[0069] For the elements in the calculated , when , it means that indirectly affects ;

[0070] For example, when the path dependency matrix is the above matrix, then:

[0071] ;

[0072] Among them, means that "Vocabulary" (v1) indirectly affects "Reading Comprehension" (v3) through "Grammar" (v2). Therefore, in subsequent knowledge point learning, if the learner masters "Vocabulary", the system can consider that "Reading Comprehension" is also indirectly mastered.

[0073] Furthermore, step S2 specifically includes the following sub-steps:

[0074] S201. Use the core knowledge points as task nodes, the dependency relationships between knowledge points as edges, and the dependency strength calculated based on the mastery degree and indirect dependency relationship as weights to establish a DTN network model;

[0075] S202. The DTN network model performs direct dependency analysis based on the direct dependency relationship and indirect dependency analysis based on the indirect dependency relationship respectively;

[0076] S203. According to the analysis results, push learning tasks in combination with the indirect dependency relationship.

[0077] Specifically, when constructing a Deep Temporal Network (DTN), the core knowledge points are first used as nodes, and the dependency relationships between each knowledge point are connected by directed edges. The strength of the dependency relationship is not only determined by the direct relationship between the knowledge points but also affected by the learner's mastery level. For example, knowledge points with a higher learner mastery level have a stronger influence and can directly assist in the learning of subsequent knowledge points. In this way, the DTN network model can dynamically represent the dependencies between knowledge points and adjust the recommendation priority of tasks based on the learner's mastery level. Further, after inputting the learner's learning data (such as test scores, learning task feedback, etc.) into the DTN network model, an analysis is first performed based on the direct dependency relationships between knowledge points. The direct dependency relationship determines whether a knowledge point needs to be mastered first before learning the next knowledge point. For example, if a learner has not mastered basic grammar points, it will be difficult to understand subsequent advanced sentence patterns. The system will also conduct a more in-depth analysis through indirect dependency relationships to identify potential weak links that the learner may have overlooked. Through these two types of analysis, the system can comprehensively identify the learner's knowledge mastery status, including potential knowledge points that the learner has not mastered but should have mastered. Finally, after analyzing the learner's learning status, the system automatically recommends the most suitable learning tasks based on the learner's mastery level and dependency relationships. When recommending tasks, the knowledge points that the learner has not mastered and the dependencies between these knowledge points and other knowledge points are considered. The learning content recommended first is those knowledge points that the learner is weak in, have a close dependency relationship, and have a moderate learning difficulty. Through the above method, the task recommendation can be dynamically adjusted to ensure that the learner can learn targeted content while avoiding repeating what has already been mastered.

[0078] Further, in step S202, the specific calculation process of direct dependency analysis is as follows: for each knowledge point, the impact of direct dependency is evaluated according to the path dependency matrix, and the learner's mastery of the directly dependent knowledge points is calculated, that is:

[0079] ;

[0080] Among them, the represents the weighted sum of the mastery levels of the knowledge points directly dependent on the target knowledge point, the represents the index of the target knowledge point, the represents the direct dependency relationship from the starting knowledge point to the target knowledge point, and the represents the mastery level of the target knowledge point.

[0081] Further, in step S202, the specific calculation process of indirect dependency analysis is as follows: for each knowledge point, calculate its mastery level through indirect path dependency, and calculate the potential indirect knowledge points mastered by the learner based on the indirect dependency relationship:

[0082] ;

[0083] Among them, the represents the mastery degree of potential knowledge points depending on k steps, and the represents the index of the power, and the represents the indirect dependency relationship between the starting knowledge point and the target knowledge point, and the represents the mastery degree of the target knowledge point.

[0084] Furthermore, the step S3 specifically includes the following sub-steps:

[0085] S301. Calculate the learning path most suitable for the current learning level of the student according to the current knowledge point mastery degree of the student and the path dependency matrix;

[0086] S302. Recommend corresponding learning content and exercises according to the learning path;

[0087] S303. Dynamically adjust the learning path according to the learning progress of the student.

[0088] Specifically, first calculate the learning path most suitable for the student according to the current knowledge mastery degree and the dependency relationship of the student. The calculation of the learning path not only depends on the knowledge mastery degree of the student, but also considers the dependency between knowledge points. Exemplarily, if the student has not mastered some basic knowledge points, the system will give priority to recommending the learning of these basic knowledge points and skip irrelevant or overly advanced content. In addition, the system will design the shortest and most efficient learning path for the student according to the priority and dependency relationship between knowledge points to ensure that the student can gradually master the knowledge system. Further, after calculating the learning path, personalized learning content will be recommended according to the path. Personalized recommendation not only recommends new knowledge for the student to learn, but also includes reviewing the content that has been mastered, especially those knowledge points that are crucial for subsequent learning. According to the current knowledge state of the student mastered by the system and the priority of the learning path, the recommended learning content will ensure that the student will not waste time on the content that has been mastered, and at the same time will not miss the core knowledge points that are crucial for subsequent learning. Finally, every time the student completes a test or a task, update the student's knowledge mastery degree according to the student's performance, and re-evaluate and optimize the learning path. When the student performs poorly on a certain knowledge point, the system will adjust the path and recommend more learning tasks related to this knowledge point; if the student has already mastered a certain content, the system will skip this part of the content and push the student into the next learning stage. Through the above dynamic optimization, it can continuously adapt to the learning progress of the student and provide a personalized learning experience.

[0089] Furthermore, in the step S301, the specific calculation process for calculating the learning path most suitable for the current learning level of the student is to calculate the path from the knowledge point reached by the current learning level to the target knowledge point for each knowledge point, that is:

[0090] ;

[0091] Among them, the represents the learning path score from the starting knowledge point to the target knowledge point, and the represents the mastery degree of the knowledge points reached by the current learning level. The represents the dependency relationship from the starting knowledge point to the knowledge points reached by the current learning level, and the represents the dependency relationship from the knowledge points reached by the current learning level to the target knowledge point.

[0092] Furthermore, it further includes step S4: After each learning cycle, train and update the DTN network model through the historical learning data of the trainees, optimize the path dependency matrix according to the learning progress and feedback of the trainees, and re-evaluate the learning status of the trainees by combining the output of the DTN network model and the dynamically adjusted path dependency matrix, and adjust the learning path and content recommendation.

[0093] Specifically, in the process of training and updating the DTN network model through the historical learning data of the trainees, the parameters of the DTN network model are updated by the gradient descent optimization algorithm.

[0094] Furthermore, after each learning cycle, collect students' learning data, including test scores, completion of learning tasks, learning time, etc. The above data helps the system update the parameters of the DTN network model to make it more accurately reflect the students' learning trends. Through historical learning data, identify which knowledge points have a greater impact on the students' learning progress and adjust the learning recommendation strategy accordingly, so as to improve the optimization accuracy of the learning path. Further, after updating the DTN network model, optimize the path dependence matrix according to the students' learning progress and feedback. The optimization of the path dependence matrix means re-evaluating the dependence relationship and dependence strength between knowledge points. For example, the learning of a certain knowledge point may become less important in some cases, or the dependence relationship between some knowledge points becomes closer. By optimizing the path dependence matrix, the system can more accurately adjust the learning path and recommended tasks, making the learning path more in line with the actual learning needs of students. Combining the updated DTN network model and the optimized path dependence matrix, re-evaluate the students' learning status. By analyzing the students' historical learning data and current learning progress, judge whether the students have mastered new knowledge points in the current learning cycle, whether there are new weak links, and adjust the learning path and recommended tasks according to this information. If a student performs outstandingly in a certain field, the system can appropriately accelerate the learning progress; if a student still has confusion in some fields, the system will extend the learning time of this part and provide more targeted learning resources. Finally, after re-evaluating the learning status, adjust the students' learning path and task recommendations to ensure that the learning effect is maximized in each learning cycle. Dynamically adjust the recommended learning content and tasks according to the students' mastery, dependence relationship and learning status. The learning path will be optimized according to the new mastery, dependence relationship and the students' feedback, and recommended learning tasks and content suitable for the current state of the students, so as to ensure that the students can receive the most appropriate learning guidance throughout the learning process.

[0095] Furthermore, as a preferred implementation method of the above embodiment, a test and evaluation system for the English learning process is proposed, including an adaptive test subsystem, a learning diagnosis subsystem and an adaptive personalized learning subsystem, where:

[0096] The adaptive test subsystem is used for students to take tests and construct a hierarchical knowledge graph based on the knowledge point dependence relationship. The hierarchical knowledge graph organizes the core knowledge points of language learning by levels and dynamically updates the mastery of each knowledge point; when the student completes a task or a test, the mastery of the knowledge points in the knowledge graph is updated in real time;

[0097] The learning diagnosis subsystem is used for diagnosing the test results of students and inputting the latest updated knowledge graph into the DTN network model. The DTN network model predicts the knowledge points mastered by the students and combines the indirect dependence relationship to identify the potential knowledge points that the students have not mastered;

[0098] The adaptive personalized learning subsystem is used to recommend the most suitable learning content and exercises for the current level of the learner and optimize the learning path by calculating the learner's knowledge mastery degree and path dependence matrix according to the intelligent diagnosis results;

[0099] Among them, the adaptive test subsystem specifically includes:

[0100] The knowledge graph module is used to initialize and construct a knowledge graph, representing the knowledge graph as a directed graph composed of a node set and an edge set, where the nodes represent core knowledge points and the edges represent the dependence relationships between knowledge points;

[0101] The dependence relationship calculation module, according to the initialized knowledge graph information, represents the direct dependence relationship between core knowledge points by constructing a path dependence matrix; obtains the indirect dependence relationship through the power operation of the matrix;

[0102] The learner test module is used to provide tests and update the mastery degree of knowledge points according to the performance of the learner in the tests and learning tasks.

[0103] Furthermore, the learning diagnosis subsystem specifically includes:

[0104] The DTN network model module is used to establish a DTN network model by taking the core knowledge points as task nodes, the dependence relationships between knowledge points as edges, and the dependence strength calculated according to the mastery degree and indirect dependence relationship as weights;

[0105] The dependence analysis module is used for the input knowledge point mastery degree, and conducts direct dependence analysis according to the direct dependence relationship and indirect dependence analysis according to the indirect dependence relationship respectively;

[0106] The result diagnosis module is used to identify the potential knowledge points that the learner has not mastered according to the analysis results.

[0107] Furthermore, the adaptive personalized learning subsystem specifically includes:

[0108] The learning path calculation module is used to calculate the most suitable learning path for the current learning level of the learner according to the current knowledge point mastery degree of the learner and the path dependence matrix;

[0109] The adaptive recommendation module is used to recommend corresponding learning content and exercises according to the learning path;

[0110] The adaptive path adjustment module dynamically adjusts the learning path according to the learning progress of the learner.

[0111] Specifically, the beneficial effects of the above implementation scheme are as follows:

[0112] Based on existing transfer learning and path learning, by applying indirect dependencies to the learning diagnosis step (S2), it is possible to diagnose and evaluate future learning;

[0113] The DTN dynamic task transfer network model is adopted, which includes a shared encoding layer and three task-specific output layers (testing, diagnosis, learning). The DTN dynamically switches task priorities according to the input knowledge graph state and the student's performance, which is more suitable for the requirements of language learning tasks.

[0114] The above are only the preferred embodiments of the present invention. It should be understood that the present invention is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments. Instead, it can be used in various other combinations, modifications, and environments, and can be changed within the scope of the concept described herein through the above teachings or the technology or knowledge in related fields. Any changes and modifications made by those skilled in the art without departing from the spirit and scope of the present invention shall fall within the protection scope of the appended claims of the present invention.

Claims

1. A test and evaluation method for an English learning process, characterized in that: The following steps are involved: S1. Construct a hierarchical knowledge graph based on knowledge point dependencies, which organizes the core knowledge points of language learning hierarchically and dynamically updates the mastery of each knowledge point; when students complete a task or test, the mastery of the knowledge points in the knowledge graph is updated in real time; S2. Input the latest updated knowledge graph into the deep temporal network model, which combines direct dependencies with indirect dependencies to identify potential knowledge points that the students have not yet mastered, and automatically determines the tasks that the students currently need to learn; S3. Based on the intelligent diagnosis results, by calculating the students’ knowledge mastery and path dependency matrix, we recommend the most suitable learning content and exercises for the students’ current level and optimize the learning path; Wherein, the step S1 specifically includes the following sub-steps: S101. Initialize and construct a knowledge graph, and represent the knowledge graph as a directed graph consisting of a node set and an edge set, where the nodes represent core knowledge points and the edges represent dependency relationships between knowledge points; S102. Based on the initialized knowledge graph information, a path dependency matrix is ​​constructed to represent the direct dependency relationship between the core knowledge points; S103. Obtaining indirect dependency through matrix exponentiation; S104. Update the mastery of knowledge points based on students’ performance in tests and learning tasks; The method comprises step S4: after each learning cycle, the deep temporal network model is trained and updated through the historical learning data of the students, and the path dependency matrix is ​​optimized according to the learning progress and feedback of the students. The output of the deep temporal network model and the dynamically adjusted path dependency matrix are combined to re-evaluate the learning status of the students, and adjust the learning path and content recommendation; The deep temporal network model includes a shared encoding layer and three task-specific output layers, and the tasks include a test task, a diagnostic task, and a learning task; The step S2 specifically includes the following sub-steps: S201. Based on the latest updated knowledge graph, the core knowledge points are used as task nodes, the dependencies between knowledge points are used as edges, and the dependency strength calculated based on the mastered knowledge points and indirect dependencies is used as weight to establish a deep temporal network model; S202. The deep temporal network model performs direct dependency analysis according to the direct dependency relationship and performs indirect dependency analysis according to the indirect dependency relationship; S203. Push learning tasks based on the analysis results and indirect dependencies.

2. A test and evaluation method for an English learning process as claimed in claim 1, characterized in that: The core knowledge point includes basic information, pre-knowledge point information and post-knowledge point information; the basic information is basic attributes, including a unique identifier, a name, and the hierarchical position of the knowledge point in the knowledge system; The pre-condition knowledge point information refers to the knowledge points that need to be mastered before learning the current knowledge point; the post-condition knowledge point information refers to other knowledge points that depend on the current knowledge point.

3. A test and evaluation method for English learning process as claimed in claim 2, characterized in that: The path dependency matrix is ​​a A matrix where each element Indicates core knowledge points Is it a core knowledge point? Prerequisite knowledge points; when knowledge points For knowledge points When the previous knowledge points are ,on the contrary .

4. A test and evaluation method for English learning process as claimed in claim 1, characterized in that: In step S103, the specific calculation process of the indirect dependency is represented by the transfer of the step dependency: ； Among them, the Represents the index of the power, represents the path dependency matrix, Indicates the number of times.

5. The test and evaluation method for English learning process as claimed in claim 1, characterized in that: In step S202, the specific calculation process of direct dependency analysis is to evaluate the impact of direct dependency on each knowledge point according to the path dependency matrix and calculate the learner's mastery of the directly dependent knowledge point, that is: ； Among them, the represents the weighted sum of the mastery of the knowledge points that the knowledge point directly depends on, represents the index of the target knowledge point, Represents the direct dependency relationship between the starting knowledge point and the target knowledge point. Indicates the mastery of the target knowledge point.

6. The method for evaluating the English learning process as claimed in claim 1, characterized in that: In step S202, the specific calculation process of the indirect dependency analysis is to calculate the mastery of each knowledge point through the indirect path dependency, and calculate the mastery of the student's potential knowledge points based on the indirect dependency relationship: ； Among them, the represents the mastery of potential knowledge points through k-step dependencies, Represents the index of the power, Represents the indirect dependency relationship from the starting knowledge point to the target knowledge point. Indicates the mastery of the target knowledge point.

7. The test and evaluation method for English learning process as claimed in claim 1, characterized in that: The step S3 specifically includes the following sub-steps: S301. Calculate the learning path that best suits the student's current learning level based on the student's current knowledge mastery and path dependency matrix; S302. Recommend corresponding learning content and exercises according to the learning path; S303. Dynamically adjust the learning path based on the students’ learning progress.

8. A test and evaluation method for English learning process as claimed in claim 7, characterized in that: In step S301, the specific calculation process of calculating the learning path that best suits the student's current learning level is to calculate the path from the knowledge point reached at the current learning level to the target knowledge point for each knowledge point, that is: ； Among them, the represents the learning path score from the starting knowledge point to the target knowledge point. Indicates the mastery of knowledge points achieved at the current learning level. Represents the dependency relationship from the starting knowledge point to the knowledge point reached at the current learning level. Represents the dependency relationship from the knowledge point reached at the current learning level to the target knowledge point. The index representing the power.

Images