Explainable methods, devices, and electronic equipment for assessing user knowledge mastery

By acquiring users' historical test information from online education systems, identifying knowledge point characteristics, and adding interpretability tags, the interpretability problem of assessing users' mastery of knowledge points is solved, improving users' understanding of learning plans and learning effectiveness.

CN116090885BActive Publication Date: 2026-11-10IFLYTEK CO LTD
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202211736173.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-12-30
Publication Date
2026-11-10
Estimated Expiration
2042-12-30

AI Technical Summary

Technical Problem

Existing online education systems lack interpretability in assessing users' mastery of knowledge points, leading to user resistance and poor results in reinforcing weak areas.

Method used

By acquiring users' historical test-taking information, we can determine the characteristics of test-taking and knowledge points. Based on a pre-defined explanation logic, we can add interpretability tags to the predicted mastery values, providing users with the accuracy of their mastery of each knowledge point and the rationale for their decisions.

Benefits of technology

It improves the transparency and accuracy of users' decision-making regarding smart education systems, enhances users' trust and acceptance of learning solutions, and improves learning outcomes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116090885B_ABST
    Figure CN116090885B_ABST
Patent Text Reader

Abstract

This invention provides an interpretable method, apparatus, and electronic device for assessing a user's mastery of knowledge points. The method includes: acquiring a user's historical test-answer information and determining the user's test-answer feature information based on the historical test-answer information; determining the knowledge point feature information corresponding to each knowledge point based on the correspondence between each test question and each knowledge point in the historical test-answer information; determining the user's predicted mastery level for each knowledge point based on the knowledge point feature information and the test-answer feature information; and adding interpretability tags to each predicted mastery level value according to a preset interpretation logic, wherein the interpretability tags are used to explain the performance of the predicted mastery level value. This method adds interpretability tags to the user's predicted mastery level value, enabling the user to understand the reasons for the intelligent education system's decisions and the accuracy of those decisions for each knowledge point.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of online education technology, and in particular to an interpretable method, apparatus, and electronic device for assessing a user's mastery of knowledge points. Background Technology

[0002] In recent years, with the development of smart education systems, more and more students are choosing online education for their studies. This is because smart education systems are very convenient; compared to traditional classroom learning, they can provide teaching resources anytime, anywhere. Secondly, smart education can utilize more data to provide more personalized and higher-quality educational resources.

[0003] Traditional online education is based on a priori methods that construct fixed learning paths. This method does not assess the user's mastery. Experienced experts develop a fixed learning plan, and each user learns step by step according to this plan. This method is just an online replica of large-class teaching, and it does not take into account the characteristics of the users themselves or ignore the differences between users.

[0004] To provide users with more personalized and targeted learning solutions, evaluating their learning progress is essential. For assessing user knowledge mastery in online education scenarios, relevant technologies include: ① Using users' historical learning records to statistically analyze their past learning progress and calculate their performance in various aspects. This method requires a large and comprehensive collection of user learning records, and its accuracy is insufficient in areas where users have limited or no records. ② Utilizing machine learning and deep learning techniques to evaluate users' mastery in various aspects. These methods are generally divided into data-driven and model-driven approaches, and their ability to characterize users is generally stronger than statistical methods.

[0005] While the above methods assess user learning, they suffer from a lack of interpretability. When a smart education system indiscriminately recommends that users strengthen their weak areas, it can easily cause user resistance and fail to achieve the goal of consolidating those weaknesses. Summary of the Invention

[0006] This invention provides an interpretable method, apparatus, and electronic device for assessing user knowledge point mastery, addressing the lack of interpretability in existing smart education systems' assessments of user knowledge point mastery. It implements an interpretable method for assessing user knowledge point mastery by adding interpretable labels to the user's predicted mastery values, thereby explaining the performance of these predicted values. This allows users to understand the rationale behind the smart education system's decisions and the accuracy of those decisions for each knowledge point.

[0007] This invention provides an interpretable method for assessing a user's mastery of knowledge points, comprising: Obtain the user's historical test answer information, and determine the user's test answer feature information based on the historical test answer information; Based on the correspondence between each question and each knowledge point in the historical test answer information, the knowledge point feature information corresponding to each knowledge point is determined; Based on the knowledge point feature information and the test question answer feature information, the predicted value of the user's mastery of each knowledge point is determined; According to the preset interpretation logic, interpretability labels are added to each of the mastery prediction values. The interpretability labels are used to interpret the performance of the mastery prediction values.

[0008] According to the present invention, an interpretable user knowledge point mastery assessment method is provided, wherein determining the knowledge point feature information corresponding to each knowledge point based on the correspondence between each question and each knowledge point in the historical test answer information includes: Based on the knowledge points marked in the historical test questions and answers, a bipartite diagram of each test question and each knowledge point is obtained. Based on the bipartite graph, the knowledge point feature information corresponding to each knowledge point is determined.

[0009] According to the present invention, an interpretable method for assessing user knowledge point mastery includes adding explanatory labels to each predicted mastery value based on a preset explanation logic, comprising: Calculate the user's mastery level for each knowledge point and the reliability of the mastery level based on the historical test answer information; Based on the mastery level and the reliability of the mastery level of each knowledge point, and the similarity between the knowledge point feature information corresponding to each knowledge point, a preset explanation logic is constructed, and an explanatory label is added to the mastery level prediction value according to the preset explanation logic.

[0010] According to the present invention, an interpretable method for evaluating user knowledge point mastery includes constructing a preset explanation logic based on the mastery level of each knowledge point, the reliability of the mastery level, and the similarity between the knowledge point feature information corresponding to each knowledge point. Based on the similarity between the feature information of each knowledge point, clustering is performed on each knowledge point to obtain multiple knowledge point clusters. Reliable knowledge points in each knowledge point cluster are selected by filtering out the credibility of the mastery level corresponding to each knowledge point. A preset explanation logic is constructed based on the mastery level of the reliable knowledge points in the knowledge point cluster and the predicted mastery level of each knowledge point in the knowledge point cluster.

[0011] According to the present invention, an interpretable method for assessing user knowledge mastery includes, before obtaining the user's historical test answer information, the method further includes: A knowledge point prediction architecture is constructed to predict a user's mastery of knowledge points. The knowledge point prediction architecture includes a graph neural network, a sequence modeling model, and a processing module. The process of obtaining the user's historical test-taking information and determining the user's test-taking feature information based on the historical test-taking information; determining the knowledge point feature information corresponding to each knowledge point based on the correspondence between each test question and each knowledge point in the historical test-taking information; and determining the user's predicted mastery level for each knowledge point based on the knowledge point feature information and the test-taking feature information includes: Obtain the user's historical test answer information, and determine the user's test answer feature information through the sequence modeling model based on the historical test answer information; Based on the correspondence between each question and each knowledge point in the historical test answer information, the knowledge point feature information corresponding to each knowledge point is determined by the graph neural network. Based on the knowledge point feature information and the test question answer feature information, the processing module determines the predicted value of the user's mastery of each knowledge point.

[0012] According to the present invention, an interpretable method for assessing user knowledge point mastery includes obtaining the user's predicted mastery value for each knowledge point through the processing module, comprising: The test question answering feature information is multiplied by the knowledge point feature information corresponding to each knowledge point to obtain the user's initial mastery prediction value for each knowledge point; The initial mastery prediction values ​​are compressed to obtain one-dimensional mastery prediction values.

[0013] The interpretable user knowledge point mastery assessment method provided by the present invention further includes: Based on the predicted mastery level for each knowledge point, the corresponding learning plan for the user is output.

[0014] The present invention also provides an interpretable device for assessing a user's mastery of knowledge points, comprising: The acquisition module is used to acquire the user's historical test answer information and determine the user's test answer feature information based on the historical test answer information; The first determining module is used to determine the knowledge point feature information corresponding to each knowledge point based on the correspondence between each question and each knowledge point in the historical test question answer information; The second determining module is used to determine the predicted value of the user's mastery of each knowledge point based on the knowledge point feature information and the test question answering feature information; An explanation module is used to add interpretability tags to each of the mastery prediction values ​​according to a preset explanation logic. The interpretability tags are used to explain the performance of the mastery prediction values.

[0015] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the interpretable user knowledge point mastery assessment method as described above.

[0016] The present invention also provides an electronic device, including a display, a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor is configured to acquire a user's historical test answer information and determine the user's test answer feature information based on the historical test answer information; determine the knowledge point feature information corresponding to each knowledge point based on the correspondence between each test question and each knowledge point in the historical test answer information; determine the user's mastery prediction value for each knowledge point based on the knowledge point feature information and the test answer feature information; add an interpretability label to each mastery prediction value according to a preset interpretation logic, the interpretability label being used to interpret the performance of the mastery prediction value; and display the user's mastery prediction value for each knowledge point and the interpretability label corresponding to each mastery prediction value.

[0017] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the interpretable user knowledge point mastery assessment method as described above.

[0018] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the interpretable user knowledge point mastery assessment method as described above.

[0019] The present invention provides an interpretable user knowledge point mastery assessment method, device, and electronic device that adds interpretability tags to the mastery prediction values ​​corresponding to each knowledge point through preset interpretation logic, so as to interpret the performance of the mastery prediction values, enabling users to understand the reasons for the smart education system's decision-making and the accuracy of the decision for each knowledge point. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating the interpretable user knowledge point mastery assessment method provided by the present invention. Figure 2 This is a flowchart illustrating the method for determining the knowledge point feature information corresponding to each of the knowledge points provided by the present invention. Figure 3 This invention provides a two-part diagram of each test question and each described knowledge point; Figure 4 This is a flowchart illustrating the method for adding explanatory tags provided by the present invention; Figure 5 This is a flowchart illustrating the method for constructing a preset interpretation logic provided by the present invention; Figure 6 This is a schematic diagram of the knowledge point prediction architecture provided by the present invention; Figure 7 This is a flowchart illustrating the method for obtaining the predicted mastery value of knowledge points provided by the present invention. Figure 8 This is a schematic diagram of the structure of the interpretable user knowledge point mastery assessment device provided by the present invention; Figure 9 This is one of the structural schematic diagrams of the electronic device provided by the present invention; Figure 10 This is the second schematic diagram of the electronic device provided by the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0023] The following is combined with Figures 1-7 The present invention describes an interpretable method for assessing user knowledge mastery, such as... Figure 1 As shown, the method includes: S101: Obtain the user's historical test answer information, and determine the user's test answer feature information based on the historical test answer information.

[0024] Specifically, users' historical test answer information can be obtained from various platforms and sources. This historical test answer information can include test content information, test difficulty, relevance to knowledge points, test quality, answer information, test score rate, and answer edge information. Among these, user answer edge information includes: answer time, test source, and knowledge points involved.

[0025] After obtaining a user's historical test answer information, the user's test answer characteristics can be determined based on this historical test answer information.

[0026] S102: Based on the correspondence between each question and each knowledge point in the historical test answer information, determine the knowledge point feature information corresponding to each knowledge point.

[0027] Specifically, based on the correspondence between each question and each knowledge point in the historical test answer information, that is, based on the knowledge points involved in each question, the knowledge point feature information corresponding to each knowledge point is determined, where one knowledge point corresponds to one knowledge point feature information.

[0028] S103: Based on the knowledge point feature information and the test question answer feature information, determine the predicted value of the user's mastery of each knowledge point.

[0029] Specifically, after determining the test question answering characteristics and knowledge point characteristics, the predicted value of the user's mastery of each knowledge point is determined based on the test question answering characteristics and knowledge point characteristics. That is, the user's mastery of the knowledge point is predicted based on the above two types of characteristics.

[0030] Based on the above, it can be understood that one user corresponds to one question's answer feature information, and one knowledge point corresponds to one knowledge point feature information. Based on the user's question-answer feature information and the corresponding knowledge point feature information, the predicted mastery level of each knowledge point can be determined. For example, based on the user's question-answer feature information and the corresponding knowledge point feature information for knowledge point A, the predicted mastery level of the user for knowledge point A can be determined; similarly, based on the user's question-answer feature information and the corresponding knowledge point feature information for knowledge point B, the predicted mastery level of the user for knowledge point B can be determined.

[0031] S104: According to the preset interpretation logic, add interpretability tags to each of the mastery prediction values. The interpretability tags are used to interpret the performance of the mastery prediction values.

[0032] Specifically, based on a preset explanation logic, interpretability labels are added to each mastery prediction value to explain its performance, such as accuracy and the rationale behind the decision. The preset explanation logic is a pre-defined explanation logic for mastery prediction values ​​under different circumstances. For example, for a mastery prediction value for knowledge point A, which is obtained based on the answer information of a relatively large number of test questions, the interpretability label for knowledge point A could be: "Predicted based on sufficient historical answer information." For a mastery prediction value for knowledge point B, which is obtained based on the answer information of a very small number of test questions, the interpretability label for knowledge point B could be: "Predicted based on historical answer information, but the historical answer information is insufficient."

[0033] As shown in the example above, based on the predicted mastery value corresponding to knowledge point A, users can know their mastery of knowledge point A. Based on the interpretability tag corresponding to knowledge point A, users can understand from the interpretability tag what the predicted mastery value for that knowledge point is based on. Furthermore, the sufficiency of historical answer information used to determine the predicted mastery value can further reveal the accuracy of the predicted mastery value. Generally speaking, the more historical answer information there is, the more accurate the predicted mastery value will be.

[0034] The present invention provides an interpretable user knowledge point mastery assessment method that adds interpretability tags to the mastery prediction values ​​corresponding to each knowledge point through preset interpretation logic, so as to interpret the performance of the mastery prediction values, enabling users to understand the reasons for the smart education system's decision-making and the accuracy of the decision for each knowledge point.

[0035] In one embodiment, such as Figure 2 As shown, determining the knowledge point feature information corresponding to each knowledge point based on the correspondence between each question and each knowledge point in the historical test answer information includes: S201: Based on the knowledge points marked in the historical test question answer information, obtain a bipartite diagram of each test question and each knowledge point.

[0036] S202: Determine the knowledge point feature information corresponding to each knowledge point based on the bipartite graph.

[0037] Specifically, based on the knowledge points marked in the historical test answer information, a bipartite diagram of each question and its corresponding knowledge points can be obtained. For example, each question has marked knowledge points, which can be single or multiple knowledge points. Figure 3 The bipartite graphs shown for each test question and each knowledge point are discretely labeled, and one test question can correspond to one or more knowledge points.

[0038] In this embodiment, by using the bipartite graphs of each test question and each knowledge point, more accurate knowledge point feature information can be obtained.

[0039] In one embodiment, such as Figure 4 As shown, adding explanatory labels to each of the predicted mastery levels according to a preset explanation logic includes: S401: Calculate the user's mastery level and the reliability of the mastery level for each knowledge point based on the historical test answer information.

[0040] S402: Construct a preset explanation logic based on the mastery level of each knowledge point, the credibility of the mastery level, and the similarity between the knowledge point feature information corresponding to each knowledge point, and add an explanatory label to the mastery level prediction value according to the preset explanation logic.

[0041] Specifically, based on historical test answer information, the user's mastery level and the reliability of that mastery level for each knowledge point can be calculated. For example, calculating knowledge points based on historical test answer information... degree of mastery And the credibility of the degree of control : in, This indicates the percentage of test scores. Indicates the difficulty level of the test questions; Indicates the degree of relevance to the knowledge point; Indicates the quality of the test questions; This indicates the total number of history questions. Indicates user; The forgetting curve function is defined by the time elapsed since the user's response. related; for The function normalizes the credibility.

[0042] After obtaining the mastery level and the credibility of each knowledge point, a pre-defined explanation logic is constructed based on the mastery level, the credibility of the mastery level, and the similarity between the knowledge point feature information corresponding to each knowledge point, and explanatory labels are added to the mastery level prediction value.

[0043] For example, a pre-defined explanation logic can be constructed based on the mastery level, the credibility of the mastery level, and the similarity between the feature information of each knowledge point. For instance, if the mastery level of knowledge point A is X, the credibility of the mastery level is Y, and the predicted mastery level is Z, when X and Y are close and the credibility Y is high, the interpretability label for knowledge point A can be: "Obtained based on historical answer predictions, and close to the mastery level calculated from historical answers, high credibility." As another example, if knowledge point A and knowledge point B are similar knowledge points (meaning they are close), and there is a significant difference between the predicted mastery level of knowledge point A and the predicted mastery level of knowledge point B, then an explanation can be provided in the interpretability label.

[0044] In this embodiment, a preset explanation logic is constructed by considering the mastery level and credibility of knowledge points, as well as the similarity between the feature information of each knowledge point. This makes the explanation content of the explanatory tags corresponding to knowledge points more comprehensive and more evidence-based, thereby improving users' trust and acceptance.

[0045] In one embodiment, such as Figure 5 As shown, the construction of a preset explanation logic based on the mastery level and the reliability of the mastery level of each knowledge point, as well as the similarity between the knowledge point feature information corresponding to each knowledge point, includes: S501: Based on the similarity between the feature information of each knowledge point, cluster the knowledge points to obtain multiple knowledge point clusters.

[0046] S502: Filter out reliable knowledge points in each knowledge point cluster based on the credibility of the mastery level corresponding to each knowledge point.

[0047] S503: Construct a preset explanation logic based on the mastery degree of the reliable knowledge points in the knowledge point cluster and the predicted mastery degree value of each knowledge point in the knowledge point cluster.

[0048] Specifically, firstly, a clustering algorithm is used to cluster knowledge points based on the similarity between their corresponding feature information, resulting in multiple knowledge point clusters. The clustering algorithm can be either hierarchical clustering or density-based clustering, or other clustering algorithms; no specific limitation is made here. Clustering knowledge points means grouping similar knowledge points into one category, forming a knowledge point cluster.

[0049] Secondly, for a knowledge point cluster, the credibility of each knowledge point in the cluster is obtained. When the credibility of a knowledge point is higher than a first preset value, the knowledge point is defined as a reliable knowledge point in the cluster. It is understandable that a knowledge point cluster may have no reliable knowledge points, or it may have one or more reliable knowledge points.

[0050] Finally, a pre-defined explanation logic is constructed based on the mastery level of reliable knowledge points within the knowledge point cluster and the predicted mastery level of each knowledge point within the cluster. For example, if knowledge point A is a reliable knowledge point with high credibility, its mastery level can be directly used instead of the predicted mastery level to output to the user. The interpretability label added to the predicted mastery level is then added accordingly. The interpretability label could be calculated based on historical answer information.

[0051] For unreliable knowledge points within a knowledge point cluster, explanatory labels can be added to the predicted mastery levels of the unreliable knowledge points based on the difference between their predicted mastery levels and those of the reliable knowledge points in the cluster. For example, if knowledge point A and knowledge point B are in the same knowledge point cluster, with A being a reliable point and B being an unreliable point, and the predicted mastery level of knowledge point B differs from that of knowledge point A by a second preset value (indicating they are relatively close), then the explanatory label for the predicted mastery level of knowledge point B could be: "Based on prediction, and similar to the mastery level calculated from historical responses of similar points." This label indicates that the predicted mastery level of knowledge point B is relatively accurate. If the predicted mastery level of knowledge point B differs from that of knowledge point A by a value exceeding the second preset value, then the predicted mastery level of knowledge point B is not accurate enough, and the corresponding explanatory label is determined.

[0052] If there are no reliable knowledge points in a knowledge point cluster, then only the predicted mastery value of each knowledge point can be considered, and the corresponding interpretability label can be determined, such as: obtained only from historical answer predictions.

[0053] In this embodiment, knowledge points are classified according to similarity, which improves the efficiency of adding interpretability tags for each knowledge point.

[0054] In one embodiment, before obtaining the user's historical test answer information, the method further includes: A knowledge point prediction architecture is constructed to predict a user's mastery of knowledge points. The knowledge point prediction architecture includes a graph neural network, a sequence modeling model, and a processing module. The process of obtaining the user's historical test-taking information and determining the user's test-taking feature information based on the historical test-taking information; determining the knowledge point feature information corresponding to each knowledge point based on the correspondence between each test question and each knowledge point in the historical test-taking information; and determining the user's predicted mastery level for each knowledge point based on the knowledge point feature information and the test-taking feature information includes: Obtain the user's historical test answer information, and determine the user's test answer feature information through the sequence modeling model based on the historical test answer information; Based on the correspondence between each question and each knowledge point in the historical test answer information, the knowledge point feature information corresponding to each knowledge point is determined by the graph neural network. Based on the knowledge point feature information and the test question answer feature information, the processing module determines the predicted value of the user's mastery of each knowledge point.

[0055] Specifically, a knowledge point prediction architecture is constructed to predict users' mastery of knowledge points. The knowledge point prediction architecture is as follows: Figure 6 As shown, it includes a graph neural network, a sequence modeling model, and a processing module.

[0056] After obtaining a user's historical test answer information, the user can be identified based on that information. The sequence of questions to answer: ,in, It is the user in The information provided in the timeline of responses. Information about the test questions answered by the user at a specific time. This refers to the user's response at a specific moment. For edge information of a user's response at a certain moment, For users The length of the question-answer sequence is the total number of historical questions. After obtaining the user's question-answer sequence, the user's question-answer feature information is determined through a sequence modeling model. The sequence modeling model can be a BERT model, and the question-answer information can be the user's question-answer vector.

[0057] Based on the correspondence between each question and each knowledge point in the historical test answer information, such as a bipartite graph illustrating the correspondence, a graph neural network can be used to obtain the association between each question and each knowledge point, thereby determining the knowledge point feature information corresponding to each knowledge point. Among them, the knowledge point feature information can be a knowledge point vector. Corresponding to knowledge point 1, Corresponding knowledge points .

[0058] It is understandable that constructing the knowledge point prediction architecture described above includes building an initial knowledge point prediction architecture and training it. During training, the objective of optimizing this knowledge point prediction architecture is to minimize the following loss function: in, This indicates the user predicted by the knowledge point prediction architecture. In the knowledge points The predicted value of mastery on the surface This indicates that the user is at the knowledge point The level of mastery obtained from the above calculation, Indicates the credibility of the degree of control. The structural risk represents the characteristic information of knowledge points, i.e., the knowledge point vector. Represents the coefficient.

[0059] In this embodiment, the knowledge point prediction architecture can quickly predict the user's mastery of each knowledge point, and the model can be reused after construction, which is convenient and fast.

[0060] In one embodiment, such as Figure 7 As shown, obtaining the user's predicted mastery level for each knowledge point through the processing module includes: S701: Multiply the test question answer feature information by the knowledge point feature information corresponding to each of the knowledge points to obtain the user's initial mastery prediction value for each of the knowledge points.

[0061] S702: Compress each of the initial mastery prediction values ​​to obtain a one-dimensional mastery prediction value.

[0062] Specifically, based on the knowledge point feature information and the test question answer feature information, when the processing module determines the user's mastery prediction value for each knowledge point, it first multiplies the test question answer feature information with the knowledge point feature information corresponding to each knowledge point to obtain an initial mastery prediction value. Then, it compresses the initial mastery prediction value to obtain a one-dimensional mastery prediction value.

[0063] For example, multiplying the question answer feature information by the knowledge point feature information corresponding to each of the aforementioned knowledge points specifically involves multiplying the question answer feature information by the knowledge point feature information of knowledge point 1. Multiplying these values ​​yields the initial predicted mastery level for knowledge point 1. The test-taking characteristic information is then compared with the knowledge point characteristic information for knowledge point 2. Multiplying these values ​​yields the initial predicted mastery level for knowledge point 2. This process is repeated until the predicted mastery level for knowledge point 2 is obtained. The initial mastery prediction value.

[0064] In one embodiment, it also includes: Based on the predicted mastery level for each knowledge point, the corresponding learning plan for the user is output.

[0065] Specifically, after determining the user's mastery prediction value and interpretability label, a corresponding learning plan can be generated and output based on the mastery prediction value and interpretability label for the user to use.

[0066] In this embodiment, a targeted learning plan is provided for the user, enabling the user to learn and improve in a targeted manner, thereby improving the user's learning effectiveness.

[0067] The following describes the explainable user knowledge point mastery assessment device provided by the present invention. The explainable user knowledge point mastery assessment device described below can be referred to in correspondence with the explainable user knowledge point mastery assessment method described above.

[0068] like Figure 8 As shown, the interpretable user knowledge point mastery assessment device includes: The acquisition module 801 is used to acquire the user's historical test answer information and determine the user's test answer feature information based on the historical test answer information; The first determining module 802 is used to determine the knowledge point feature information corresponding to each knowledge point based on the correspondence between each question and each knowledge point in the historical test question answer information; The second determining module 803 is used to determine the predicted value of the user's mastery of each knowledge point based on the knowledge point feature information and the test question answering feature information; The explanation module 804 is used to add interpretability tags to each of the mastery prediction values ​​according to a preset explanation logic. The interpretability tags are used to explain the performance of the mastery prediction values.

[0069] In one embodiment, the first determining module 802 is specifically used for: Based on the knowledge points marked in the historical test questions and answers, a bipartite diagram of each test question and each knowledge point is obtained. Based on the bipartite graph, the knowledge point feature information corresponding to each knowledge point is determined.

[0070] In one embodiment, the interpretation module 804 is specifically used for: Calculate the user's mastery level for each knowledge point and the reliability of the mastery level based on the historical test answer information; Based on the mastery level and the reliability of the mastery level of each knowledge point, and the similarity between the knowledge point feature information corresponding to each knowledge point, a preset explanation logic is constructed, and an explanatory label is added to the mastery level prediction value according to the preset explanation logic.

[0071] In one embodiment, the interpretation module 804 is further configured to: Based on the similarity between the feature information of each knowledge point, clustering is performed on each knowledge point to obtain multiple knowledge point clusters. Reliable knowledge points in each knowledge point cluster are selected by filtering out the credibility of the mastery level corresponding to each knowledge point. A preset explanation logic is constructed based on the mastery level of the reliable knowledge points in the knowledge point cluster and the predicted mastery level of each knowledge point in the knowledge point cluster.

[0072] In one embodiment, the explainable user knowledge point mastery assessment device further includes: A construction module is used to build a knowledge point prediction architecture for predicting a user's mastery of knowledge points. The knowledge point prediction architecture includes a graph neural network, a sequence modeling model, and a processing module. The acquisition module 801, the first determination module 802, and the second determination module 803 are specifically used for: Obtain the user's historical test answer information, and determine the user's test answer feature information through the sequence modeling model based on the historical test answer information; Based on the correspondence between each question and each knowledge point in the historical test answer information, the knowledge point feature information corresponding to each knowledge point is determined by the graph neural network. Based on the knowledge point feature information and the test question answer feature information, the processing module determines the predicted value of the user's mastery of each knowledge point.

[0073] In one embodiment, the second determining module 803 is further configured to: The test question answering feature information is multiplied by the knowledge point feature information corresponding to each knowledge point to obtain the user's initial mastery prediction value for each knowledge point; The initial mastery prediction values ​​are compressed to obtain one-dimensional mastery prediction values.

[0074] In one embodiment, the explainable user knowledge point mastery assessment device further includes: The learning plan output module is used to output the learning plan corresponding to the user based on the mastery prediction value corresponding to each knowledge point.

[0075] The interpretable user knowledge point mastery assessment device provided by this invention adds interpretability tags to the mastery prediction values ​​corresponding to each knowledge point through preset interpretation logic, so as to interpret the performance of the mastery prediction values, enabling users to understand the reasons for the smart education system's decision-making and the accuracy of the decision for each knowledge point.

[0076] Figure 9 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 9As shown, the electronic device may include: a processor 910, a communication interface 920, a memory 930, and a communication bus 940, wherein the processor 910, the communication interface 920, and the memory 930 communicate with each other through the communication bus 940. The processor 910 can call logical instructions in the memory 930 to execute an interpretable user knowledge point mastery assessment method. This method includes: acquiring the user's historical test answer information and determining the user's test answer feature information based on the historical test answer information; determining the knowledge point feature information corresponding to each knowledge point based on the correspondence between each test question and each knowledge point in the historical test answer information; determining the user's predicted mastery value for each knowledge point based on the knowledge point feature information and the test answer feature information; and adding an interpretability tag to each predicted mastery value according to preset interpretation logic, wherein the interpretability tag is used to interpret the performance of the predicted mastery value.

[0077] Furthermore, the logical instructions in the aforementioned memory 930 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0078] Figure 10 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 10As shown, the electronic device includes a display 1050, a memory 1030, a processor 1010, a computer program stored in the memory 1030 and executable on the processor, a communication interface 1020, and a communication bus 1040. The processor 1010, the communication interface 1020, the memory 1030, and the display 1050 communicate with each other through the communication bus 1040. The processor 1010 is used to acquire the user's historical test answer information, and determine the user's test answer feature information based on the historical test answer information; determine the knowledge point feature information corresponding to each knowledge point based on the correspondence between each test question and each knowledge point in the historical test answer information; determine the user's mastery prediction value for each knowledge point based on the knowledge point feature information and the test answer feature information; add interpretability tags to each mastery prediction value according to preset explanation logic, the interpretability tags being used to explain the performance of the mastery prediction value; the display 1050 is used to display the user's mastery prediction value for each knowledge point and the interpretability tags corresponding to each mastery prediction value.

[0079] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the interpretable user knowledge point mastery assessment method provided by the above methods. The method includes: acquiring the user's historical test question answer information and determining the user's test question answer feature information based on the historical test question answer information; determining the knowledge point feature information corresponding to each knowledge point based on the correspondence between each test question and each knowledge point in the historical test question answer information; determining the user's mastery prediction value for each knowledge point based on the knowledge point feature information and the test question answer feature information; and adding an interpretability label to each mastery prediction value according to a preset interpretation logic. The interpretability label is used to interpret the performance of the mastery prediction value.

[0080] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the interpretable user knowledge point mastery assessment method provided by the methods described above. This method includes: acquiring a user's historical test-answer information and determining the user's test-answer feature information based on the historical test-answer information; determining knowledge point feature information corresponding to each knowledge point based on the correspondence between each test question and each knowledge point in the historical test-answer information; determining the user's predicted mastery level for each knowledge point based on the knowledge point feature information and the test-answer feature information; and adding an interpretability tag to each predicted mastery level according to a preset interpretation logic, wherein the interpretability tag is used to interpret the performance of the predicted mastery level.

[0081] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0082] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0083] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An interpretable method for assessing users' mastery of knowledge points, characterized in that, include: A knowledge point prediction architecture is constructed to predict a user's mastery of knowledge points. The knowledge point prediction architecture includes a graph neural network, a sequence modeling model, and a processing module. Obtain the user's historical test answer information, and determine the user's test answer feature information through the sequence modeling model based on the historical test answer information; Based on the correspondence between each question and each knowledge point in the historical test answer information, the knowledge point feature information corresponding to each knowledge point is determined by the graph neural network. Based on the knowledge point feature information and the test question answer feature information, the processing module determines the predicted value of the user's mastery of each knowledge point. Calculate the user's mastery level for each knowledge point and the reliability of the mastery level based on the historical test answer information; Based on the similarity between the feature information of each knowledge point, clustering is performed on each knowledge point to obtain multiple knowledge point clusters. Reliable and unreliable knowledge points in each knowledge point cluster are filtered out by the credibility of the mastery level corresponding to each knowledge point, and the credibility of the reliable knowledge points is higher than the first preset value. A preset explanation logic is constructed based on the mastery degree of the reliable knowledge points in the knowledge point cluster and the predicted mastery degree of each knowledge point in the knowledge point cluster. According to the preset interpretation logic, an interpretability label is added to each of the mastery prediction values. The interpretability label is used to interpret the performance of the mastery prediction value. Specifically, explanatory labels are added to the predicted mastery values ​​of unreliable knowledge points based on the difference between the predicted mastery values ​​of unreliable knowledge points and the mastery values ​​of reliable knowledge points in the cluster. Mastery of the knowledge points It is calculated using the following formula: ; The credibility of the stated level of mastery It is calculated using the following formula: ; in, This represents the score rate of the t-th question; This indicates the difficulty level of the t-th question. This represents the degree of correlation between the t-th question and the k-th knowledge point; This represents the quality of the t-th question. This indicates the total number of history questions. Indicates user; The forgetting curve function is defined by the time elapsed since the user's response. related; for The function normalizes the credibility.

2. The interpretable user knowledge point mastery assessment method according to claim 1, characterized in that, The step of determining the knowledge point feature information corresponding to each knowledge point through the graph neural network based on the correspondence between each question and each knowledge point in the historical test answer information includes: Based on the knowledge points marked in the historical test questions and answers, a bipartite diagram of each test question and each knowledge point is obtained. Based on the bipartite graph, the graph neural network is used to determine the knowledge point feature information corresponding to each knowledge point.

3. The interpretable user knowledge point mastery assessment method according to claim 1 or 2, characterized in that, The step of obtaining the user's predicted mastery level for each knowledge point through the processing module includes: The test question answering feature information is multiplied by the knowledge point feature information corresponding to each knowledge point to obtain the user's initial mastery prediction value for each knowledge point; The initial mastery prediction values ​​are compressed to obtain one-dimensional mastery prediction values.

4. The interpretable user knowledge point mastery assessment method according to claim 1 or 2, characterized in that, Also includes: Based on the predicted mastery level for each knowledge point, the corresponding learning plan for the user is output.

5. An interpretable device for assessing user knowledge mastery, characterized in that, include: The acquisition module is used to construct a knowledge point prediction architecture for predicting a user's mastery of knowledge points. The knowledge point prediction architecture includes a graph neural network, a sequence modeling model, and a processing module. Obtain the user's historical test answer information, and determine the user's test answer feature information through the sequence modeling model based on the historical test answer information; The first determining module is used to determine the knowledge point feature information corresponding to each knowledge point by means of the graph neural network based on the correspondence between each question and each knowledge point in the historical question answer information. The second determining module is used to determine the predicted value of the user's mastery of each knowledge point based on the knowledge point feature information and the test question answer feature information, through the processing module. An explanation module is used to calculate the user's mastery level for each knowledge point and the reliability of the mastery level based on the historical test answer information; Based on the similarity between the feature information of each knowledge point, clustering is performed on each knowledge point to obtain multiple knowledge point clusters. Reliable and unreliable knowledge points in each knowledge point cluster are filtered out by the credibility of the mastery level corresponding to each knowledge point, and the credibility of the reliable knowledge points is higher than the first preset value. A preset explanation logic is constructed based on the mastery degree of the reliable knowledge points in the knowledge point cluster and the predicted mastery degree of each knowledge point in the knowledge point cluster. According to the preset interpretation logic, an interpretability label is added to each of the mastery prediction values. The interpretability label is used to interpret the performance of the mastery prediction value. Specifically, explanatory labels are added to the predicted mastery values ​​of unreliable knowledge points based on the difference between the predicted mastery values ​​of unreliable knowledge points and the mastery values ​​of reliable knowledge points in the cluster. Mastery of the knowledge points It is calculated using the following formula: ; The credibility of the stated level of mastery It is calculated using the following formula: ; in, This represents the score rate of the t-th question; This indicates the difficulty level of the t-th question. This represents the degree of correlation between the t-th question and the k-th knowledge point; This represents the quality of the t-th question. This indicates the total number of history questions. Indicates user; The forgetting curve function is defined by the time elapsed since the user's response. related; for The function normalizes the credibility.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the interpretable user knowledge point mastery assessment method as described in any one of claims 1 to 4.

7. An electronic device comprising a display, a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor is used to construct a knowledge point prediction architecture for predicting a user's mastery of knowledge points. The knowledge point prediction architecture includes a graph neural network, a sequence modeling model, and a processing module. Obtain the user's historical test answer information, and determine the user's test answer feature information through the sequence modeling model based on the historical test answer information; Based on the correspondence between each question and each knowledge point in the historical test answer information, the knowledge point feature information corresponding to each knowledge point is determined by the graph neural network. Based on the knowledge point feature information and the test question answer feature information, the processing module determines the predicted value of the user's mastery of each knowledge point; and calculates the user's mastery of each knowledge point and the reliability of the mastery based on the historical test question answer information. Based on the similarity between the feature information of each knowledge point, clustering is performed on each knowledge point to obtain multiple knowledge point clusters. Reliable and unreliable knowledge points in each knowledge point cluster are filtered out by the credibility of the mastery level corresponding to each knowledge point, and the credibility of the reliable knowledge points is higher than the first preset value. A preset explanation logic is constructed based on the mastery degree of the reliable knowledge points in the knowledge point cluster and the predicted mastery degree of each knowledge point in the knowledge point cluster. According to the preset interpretation logic, an interpretability label is added to each of the mastery prediction values. The interpretability label is used to interpret the performance of the mastery prediction value. The display is used to show the user's predicted mastery level for each of the knowledge points and the interpretability label corresponding to each predicted mastery level. Specifically, explanatory labels are added to the predicted mastery values ​​of unreliable knowledge points based on the difference between the predicted mastery values ​​of unreliable knowledge points and the mastery values ​​of reliable knowledge points in the cluster. Mastery of the knowledge points It is calculated using the following formula: ; The credibility of the stated level of mastery It is calculated using the following formula: ; in, This represents the score rate of the t-th question; This indicates the difficulty level of the t-th question. This represents the degree of correlation between the t-th question and the k-th knowledge point; This represents the quality of the t-th question. This indicates the total number of history questions. Indicates user; The forgetting curve function is defined by the time elapsed since the user's response. related; for The function normalizes the credibility.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the interpretable user knowledge point mastery assessment method as described in any one of claims 1 to 4.

Citation Information

Patent Citations

  • Learning condition diagnosis method, device and equipment and storage medium

    CN112053269A

  • Student portrait construction and learning resource distribution method, computer equipment and storage medium

    CN114357297A

  • Explanatable knowledge cognition level mining method and system fusing learning cognition elements

    CN115422256A