Method and device for measuring knowledge content learning result in logarithmic intelligent courseware

Abstract

The invention provides a method and a device for measuring a knowledge content learning result in digital intelligent courseware. The problem that the knowledge content measurement learning result of a student cannot be accurately judged based on the digital intelligent courseware is effectively solved. The method comprises the following steps: configuring an annotation attribute used for measuring a first digital intelligent courseware learning result, and obtaining multi-type learning data under the annotation attribute generated when a student learns the first digital intelligent courseware in a preset historical time period; obtaining target learning data under the key annotation attribute; constructing and training a plurality of classification models based on the target learning data, and determining a target classification model from the plurality of trained classification models; acquiring to-be-tested learning data of a to-be-tested student for a second digital intelligent courseware, inputting the to-be-tested learning data into the target classification model to output a classification result, and measuring a learning result of the to-be-tested student for knowledge content in the second digital intelligent courseware based on the classification result.

Description

Technical Field

[0001] The present application relates to the field of educational assessment technology, and more specifically, to a method and device for measuring learning outcomes of knowledge content in digital intelligent courseware. Background Art

[0002] Intelligent Courseware is an advanced teaching resource that integrates digital and intelligent technologies. It is built on digital technology and converts traditional textbooks, teaching materials and various educational resources into digital forms that can be processed by computers, which is convenient for storage, management and dissemination. Intelligent Courseware uses multimedia, animation, simulation and other means to provide a highly interactive learning experience, enabling students to understand and master knowledge in a dynamic and visual way, and enhance participation and fun in the learning process. In addition, intelligent courseware combines artificial intelligence and big data analysis technology to dynamically adjust and personalize according to learners' personal behavior, ability level and personalized needs to achieve adaptive teaching. Intelligent courseware also has real-time learning data collection and analysis functions, provides feedback and evaluation, helps course designers optimize teaching strategies, and guides students to improve their learning methods. In addition, intelligent courseware supports students' autonomous learning and exploratory learning, and cultivates students' innovative thinking ability and independent problem-solving ability. Intelligent courseware aims to empower education through technology, improve teaching quality, promote efficient learning, and meet the development trend of modern education informatization and personalized teaching.

[0003] However, in the education evaluation model based on digital courseware, although more emphasis is placed on the memory and understanding levels, there are deficiencies in paying attention to students' deep understanding and application capabilities of knowledge concepts; there is also a lack of sufficient attention to individual differences among students, as well as customized feedback on their learning paths; the volume of educational data generated by digital courseware is usually large and complex, and manual processing and analysis of data is not only time-consuming and inefficient, but digital courseware cannot respond in real time and adjust teaching strategies according to the level of students' knowledge mastery, and cannot respond to changes in students' learning status and updates in educational needs, and has a low accuracy rate. Summary of the invention

[0004] In view of this, the purpose of this application is to provide a method and device for measuring the learning results of knowledge content in digital courseware. This method and device for measuring the learning results of knowledge content in digital courseware effectively solves the problem that it is impossible to accurately judge the students' knowledge content measurement learning results based on digital courseware.

[0005] In a first aspect, an embodiment of the present application provides a method for measuring the learning results of knowledge content in a digital intelligent courseware, the method comprising:

[0006] Configuring a labeling attribute for measuring a learning result of the first digital courseware, and obtaining multiple types of learning data under the labeling attribute generated by students learning the first digital courseware within a preset historical time period;

[0007] Calculating the influence coefficient of the annotation attribute corresponding to the multi-type learning data on the measurement learning result, and acquiring the target learning data under the key annotation attribute from the multi-type learning data based on the influence coefficient;

[0008] Constructing and training multiple classification models based on the target learning data, and determining a target classification model from the trained multiple classification models;

[0009] The learning data to be tested of the student to be tested for the second digital courseware is obtained, and the learning data to be tested is input into the target classification model to output a classification result, and the learning result of the student to be tested for the knowledge content in the second digital courseware is measured based on the classification result.

[0010] In combination with the first aspect, the embodiment of the present application provides a first possible implementation of the first aspect, wherein the multi-type learning data under the annotated attributes generated by the student learning the first digital courseware within a preset historical time period includes:

[0011] Generate a plurality of attribute fields based on pre-configured annotation attributes, and acquire and store corresponding learning data according to the attribute fields to construct a first database;

[0012] Acquire, from the first database, multiple types of learning data generated by students for the first digital courseware;

[0013] The obtaining of target learning data under key annotation attributes includes:

[0014] Acquire target learning data under key annotation attributes from the first database;

[0015] A second database is constructed based on the target learning data, so as to construct and train multiple classification models based on the target learning data in the second database.

[0016] In combination with the first aspect, the embodiment of the present application provides a second possible implementation of the first aspect, wherein the obtaining of the multi-type learning data under the annotated attributes generated by the student learning the first digital courseware within a preset historical time period includes:

[0017] Determine a plurality of attribute fields corresponding to the annotation attribute;

[0018] Matching is performed according to multiple data attributes corresponding to the attribute field to obtain multiple types of learning data matching the attribute field.

[0019] In combination with the first aspect, the embodiment of the present application provides a third possible implementation manner of the first aspect, wherein the determining the multiple attribute fields corresponding to the annotation attribute includes:

[0020] Obtaining a mapping relationship between the annotation attribute and the attribute field;

[0021] According to the mapping relationship, a plurality of attribute fields corresponding to the annotation attribute are determined.

[0022] In combination with the first aspect, an embodiment of the present application provides a fourth possible implementation of the first aspect, wherein the obtaining, based on the influence coefficient, target learning data under the key annotated attribute from the learning data with the annotated attribute includes:

[0023] Calculating the influence coefficient of the annotation attribute on the student's measured learning results based on the computing network model, and judging whether the influence coefficient meets the preset correlation condition;

[0024] If so, the corresponding annotation attribute is determined to be the key annotation attribute, and the corresponding learning data is determined to be the target learning data.

[0025] In combination with the first aspect, the embodiment of the present application provides a fifth possible implementation of the first aspect, wherein determining the target classification model from the trained multiple classification models includes:

[0026] Evaluate the multiple classification models respectively based on multiple evaluation dimensions to obtain evaluation results corresponding to each classification model;

[0027] The best evaluation result is selected from the multiple evaluation results, and the classification model corresponding to the best evaluation result is determined as the target classification model.

[0028] In combination with the first aspect, the embodiment of the present application provides a sixth possible implementation of the first aspect, wherein the multiple classification models are calculated separately based on multiple evaluation dimensions to obtain an evaluation result corresponding to each classification model, including:

[0029] Determine multiple evaluation dimensions and test data sets for multiple trained classification models;

[0030] Based on the test data set, the multiple classification models are tested from multiple evaluation dimensions to obtain corresponding test results, and based on the test results, they are evaluated to obtain corresponding evaluation results.

[0031] In a second aspect, an embodiment of the present application provides a device for measuring the learning results of knowledge content in a digital courseware, the device comprising:

[0032] A configuration module, used to configure a labeling attribute for measuring a learning result of the first digital courseware, and to obtain multiple types of learning data under the labeling attribute generated by students learning the first digital courseware within a preset historical time period;

[0033] A calculation module, used to calculate the influence coefficient of the annotation attribute corresponding to the multi-type learning data on the measurement learning result, and obtain the target learning data under the key annotation attribute from the multi-type learning data based on the influence coefficient;

[0034] A training module, used to construct and train multiple classification models based on the target learning data, and determine a target classification model from the trained multiple classification models;

[0035] The measurement module is used to obtain the learning data to be tested of the students for the second digital courseware, input the learning data to be tested into the target classification model to output the classification result, and measure the learning results of the students to be tested for the knowledge content in the second digital courseware based on the classification result.

[0036] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the memory through the bus, and when the machine-readable instructions are executed by the processor, the steps of any one of the methods for measuring the learning outcomes of knowledge content in digital intelligent courseware are performed.

[0037] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the computer program executes the steps of any one of the methods for measuring the learning outcomes of knowledge content in digital intelligent courseware.

[0038] An embodiment of the present application provides a method for measuring the learning results of knowledge content in a digital courseware. The method first configures a labeling attribute for measuring the learning results of a first digital courseware, and obtains multi-type learning data under the labeling attribute generated by students learning the first digital courseware within a preset historical time period; secondly, calculates the influence coefficient of the labeling attribute corresponding to the multi-type learning data on the measured learning results, and obtains target learning data under key labeling attributes from the multi-type learning data based on the influence coefficient; then, constructs and trains multiple classification models based on the target learning data, and determines the target classification model from the trained multiple classification models; finally, obtains the learning data to be tested of the students for the second digital courseware, and inputs the learning data to be tested into the target classification model to output the classification result, and measures the learning data to be tested based on the classification result. The students' learning results of the knowledge content in the second digital courseware can accurately judge the measurement learning results of the students' knowledge content in the digital courseware, thereby paying attention to the students' deep understanding and application ability of knowledge concepts, and determining how to respond to changes in students' learning status and updates in educational needs based on their knowledge results. Based on the training model, a large amount of educational data can be automatically processed and analyzed, data processing efficiency can be improved, and errors can be reduced. This application analyzes individual students' learning activity data and grades, combines machine learning for accurate data interpretation, and can provide targeted personalized learning suggestions and feedback, promote students' personalized development, and also contribute to a more reasonable allocation of educational resources. Educators can obtain timely measurement learning results and dynamically adjust teaching content and methods to improve teaching effectiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0040] Figure 1 A schematic diagram of a process of measuring the learning results of knowledge content in digital intelligent courseware provided in an embodiment of the present application is shown;

[0041] Figure 2 A schematic diagram of a decision tree classification model provided in an embodiment of the present application is shown;

[0042] Figure 3 A schematic diagram showing the relationship between the second database provided in an embodiment of the present application and the multiple classification models;

[0043] Figure 4A schematic diagram of the structure of a first device for measuring the learning results of knowledge content in digital intelligent courseware provided in an embodiment of the present application is shown;

[0044] Figure 5 A schematic structural diagram of a first electronic device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0045] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the drawings in the present application only serve the purpose of explanation and description and are not used to limit the scope of protection of the present application. In addition, it should be understood that the schematic drawings are not drawn in real proportion. The flowchart used in this application shows the operations implemented according to some embodiments of the present application. It should be understood that the operations of the flowchart can be implemented out of sequence, and the steps without logical context can be reversed in order or implemented simultaneously. In addition, those skilled in the art can add one or more other operations to the flowchart under the guidance of the content of the present application, or remove one or more operations from the flowchart.

[0046] In addition, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0047] It should be noted that the term "comprising" will be used in the embodiments of the present application to indicate the existence of the features declared thereafter, but does not exclude the addition of other features.

[0048] Although the current education assessment model based on digital courseware pays more attention to the memory and understanding levels, it is insufficient in paying attention to students' deep understanding and application capabilities of knowledge concepts; there is also a lack of sufficient attention to individual differences among students, as well as customized feedback on their learning paths; the volume of educational data generated by digital courseware is usually large and complex, and manual processing and analysis of data is not only time-consuming and inefficient, but also digital courseware cannot respond to and adjust teaching strategies in real time according to the level of students' knowledge mastery, and cannot respond to changes in students' learning status and updates in educational needs, and has a low accuracy rate.

[0049] Based on this, the embodiments of the present application provide a method and device for measuring the learning results of knowledge content in digital courseware, which is described below through embodiments.

[0050] Example 1

[0051] To facilitate understanding of this embodiment, firstly, a method for measuring the learning results of knowledge content in digital intelligent courseware disclosed in the embodiment of this application is introduced in detail. Figure 1 The flowchart of a method for measuring the learning results of knowledge content in digital intelligent courseware is shown. The present application provides a method for measuring the learning results of knowledge content in digital intelligent courseware, and the method includes:

[0052] S101, configuring a labeling attribute for measuring a learning result of a first digital courseware, and obtaining multiple types of learning data under the labeling attribute generated by students learning the first digital courseware within a preset historical time period;

[0053] S102, calculating the influence coefficient of the annotation attribute corresponding to the multi-type learning data on the measurement learning result, and obtaining the target learning data under the key annotation attribute from the multi-type learning data based on the influence coefficient;

[0054] S103, constructing and training multiple classification models based on the target learning data, and determining a target classification model from the trained multiple classification models;

[0055] S104. Obtain the learning data of the student to be tested for the second digital courseware, input the learning data to be tested into the target classification model to output a classification result, and measure the learning result of the student to be tested for the knowledge content in the second digital courseware based on the classification result.

[0056] In step S101, the first digital courseware is generated in the digital learning platform used by students when conducting digital learning. It can be stored in the courseware library in the digital learning platform or uploaded by teaching staff. The students described in this application can be students in a school or social workers who need to use the platform to achieve skill training. The digital learning platform can collect and store multiple types of learning data generated by students during learning, and configure the digital learning courseware to configure annotation attributes for measuring the learning results of the first digital courseware. The annotation attributes include multiple types, and the annotation attributes are PEG, SPR, STR, SCG and STG, respectively. The learning time of the target / target learning object and the comparable learning object are measured by the two abbreviations STG and STR, respectively, and the test scores obtained by students for the learning materials related to the target object and for the specific target object belong to SPR and PEG, respectively. EG, the number of times a student repeats a single target learning item to understand it is called SCG, and UNS is the name of the attribute that represents the user's knowledge level, where the user's knowledge level is also the measured learning outcome, and the attribute type to which the labeled attribute belongs is also divided into personal behavior, test scores and learning level, and obtain multi-type learning data under the labeled attribute within a preset time period, and the preset time period is the historical time period, where the multi-type learning data includes data corresponding to learning time, number of interactions, interaction frequency, test scores, accuracy, etc., and learning time, number of interactions, interaction frequency, test scores, accuracy, etc. are data attributes for the type of learning data, then obtain the multi-type learning data generated by the students for the digital courseware, determine the labeled attributes corresponding to the multi-type learning data, and obtain the multi-type learning data under the labeled attributes, so as to better understand the students' learning and mastery of knowledge based on the multi-type learning data under the labeled attributes.

[0057] In the specific implementation process of step S101, there is an embodiment in which: the obtaining of the multi-type learning data under the annotated attributes generated by the student learning the first digital courseware within a preset historical time period includes:

[0058] S10111, generating a plurality of attribute fields based on pre-configured annotation attributes, and acquiring and storing corresponding learning data according to the attribute fields to construct a first database;

[0059] S10112. Acquire, from the first database, multiple types of learning data generated by students for the first digital courseware;

[0060] The obtaining of target learning data under key annotation attributes includes:

[0061] S10113. Acquire target learning data under key annotation attributes from the first database;

[0062] S10114. Construct a second database based on the target learning data, so as to construct and train multiple classification models based on the target learning data in the second database.

[0063] In steps S10111-S10114, for the multi-type learning data generated in the digital courseware within a preset time period, the corresponding attribute fields are determined according to the data attributes of the multi-type learning data, that is, it is judged whether there is correlation and dependency between the data attributes, that is, the attribute fields corresponding to multiple data attributes are the same, then the multi-type learning data corresponding to the data attributes are stored under the same attribute field, and the corresponding labeled attributes are obtained based on the attribute fields, that is, there are multiple attribute fields under the labeled attributes, the attribute fields correspond to multiple data attributes, and the data attributes correspond to multi-type learning data, then the corresponding multi-type learning data can be obtained based on the labeled attributes, and so on, until all the multi-type learning data are stored, that is, a first database for storing multi-type learning data based on the labeled attributes is constructed; the first database has a huge amount of data, and there is inaccuracy when it is used to construct multiple classification models.

[0064] To avoid this phenomenon, after obtaining the first database, the key annotation attributes are screened out from the multiple annotation attributes in the first database, and the existence of the target learning data under the key annotation attributes is also determined. At this time, the target learning data under the key annotation attributes are obtained from the first database, that is, not all data attributes are selected, such as the correlation between the students' knowledge level UNS and the time STR they spend on relevant learning materials is the smallest. The key annotation attributes include PEG, SPR, SCG and STG respectively. STR does not belong to the data attributes for constructing the second database, so the learning data corresponding to this data attribute is not obtained and is not stored in the second database. The second database is constructed based on the target learning data corresponding to PEG, SPR, SCG and STG, so as to construct and train multiple classification models based on the target learning data in the second database. Using the second database to construct and train multiple classification models can reduce noise and redundant information, thereby improving the training efficiency and classification accuracy of multiple classification models, in order to obtain better model performance and higher measurement accuracy. In addition, the use of the second database can also reduce the consumption of computing resources, making the model training process more efficient. This method helps to improve the interpretability of the model and ensure that multiple classification models are learned and predicted only based on the most relevant features.

[0065] In the specific implementation process of step S101, there is another embodiment: the obtaining of the multi-type learning data under the annotated attributes generated by the student learning the first digital courseware within a preset historical time period includes:

[0066] S10121. Determine a plurality of attribute fields corresponding to the annotation attribute;

[0067] S10122. Match multiple data attributes corresponding to the attribute field to obtain multiple types of learning data matching the attribute field.

[0068] In steps S10121-S10122, in order to further determine the relationship between the multiple types of learning data based on the labeled attributes, the labeled attributes are divided into corresponding attribute types. The labeled attributes are pre-divided into corresponding attribute types. The attribute types include two types related to personal behavior and test scores. Each attribute type has multiple labeled attributes, among which STR, SCG and STG belong to the attribute type of personal behavior, PEG and SPR belong to the attribute type of test scores, and UNS is the attribute type of knowledge level, as shown in Table 1. Based on the labeled attribute, the corresponding attribute field is determined, that is, one labeled attribute can correspond to multiple attribute fields. After determining the attribute field, the multiple data attributes corresponding to the attribute field are further determined, and then matching is performed based on the data attribute, that is, the data attribute is matched with the data attribute of the multi-type learning data. It is necessary to pre-extract the data attributes of the multi-type learning data, such as learning time, number of interactions, interaction frequency, test scores, and accuracy. For example, the data attribute of learning time is the behavior generated by an individual during the learning process, and the data attribute of test scores and the data attribute of accuracy are the learning content results generated by the test. Then, the multi-type learning data under the labeled attributes are determined to correspond to the labeled attributes according to the data type, so as to obtain the multi-type learning data under the labeled attributes.

[0069] Table 1 Label attribute table

[0070] Property Abbreviation Property Name Property Type Attribute Description STG The learning time of the target audience Personal Behavior A measure of the time students spend on a target learning object SCG Target object's learning level count Personal Behavior The measure of the number of repetitions of the target learning object STR The learning time of the relevant subjects Personal Behavior A measure of the time students spend studying objects related to the target object SPR The learning percentage of related objects Test score related The user's score for objects related to the target object PEG Performance on the target test Test score related The user's score for the target object UNS User knowledge status Knowledge level User's knowledge level

[0071] In the specific implementation process of step S10122, there is an embodiment in which: the determining of the multiple attribute fields corresponding to the annotation attribute includes:

[0072] S101221. Obtain a mapping relationship between the annotation attribute and the attribute field;

[0073] S101222. Determine a plurality of attribute fields corresponding to the annotation attribute according to the mapping relationship.

[0074] In steps S101221-S101222, after obtaining the annotation attribute, the attribute type to which the annotation attribute belongs is determined according to the preset belonging relationship, wherein the annotation attribute is obtained by mapping the attribute field, that is, there is a corresponding mapping relationship between the annotation attribute and the attribute field, and the mapping relationship is pre-acquired to map the attribute field based on the mapping relationship to obtain the corresponding annotation attribute, the annotation attribute corresponds to multiple attribute fields, each attribute field corresponds to a data attribute, and after determining the attribute type to which it belongs, the mapping is reversed according to the preset mapping relationship to obtain multiple attribute fields corresponding to the annotation attribute, and after obtaining the attribute field, the data attribute corresponding to the attribute field and the corresponding multi-type learning data can be determined or obtained, and if there are evaluation data of others such as teacher's evaluation in the first database, then correspondingly, the corresponding data attribute of the evaluation data of others and the attribute field are determined to correspond to the annotation attribute, so as to determine the multi-type learning data under the specific annotation attribute, and also realize the effective management of the multi-type learning data based on the annotation attribute.

[0075] In step S102, after obtaining multi-type learning data under the labeled attributes, the influence coefficient of the labeled attributes on the measured learning results is calculated based on the computational network model, wherein the computational network model is determined based on the linear relationship between the influence coefficient of the labeled attributes and the measured learning results, and based on the correlation between the labeled attributes and the measured learning results, the computational network model for calculating the influence coefficient of the labeled attributes and the measured learning results is selected and determined. In the present application, the computational network model is implemented using the Pearson correlation coefficient PCC, which is the ratio between the covariance of two variables and the product of their standard deviations; therefore, it is actually a standardized measure of the covariance, and its result is always between -1 and 1. Like the covariance itself, this measurement can only reflect the linear correlation between variables and ignores many other types of relationships or correlations. As a simple example, one would expect the Pearson correlation coefficient between the age and height of a sample of teenagers in a high school to be significantly greater than 0, but less than 1 (because 1 would represent an unrealistic perfect correlation). After selecting the Pearson correlation coefficient PCC as the computational network model, each annotation attribute is processed by the computational network model to obtain the corresponding influence coefficient. Based on the influence coefficient, the key annotation attribute is determined, and based on the influence coefficient of the annotation attribute in the learning data on the measured learning results, the key annotation attributes are screened from the annotation attributes, that is, the annotation attributes that have a greater impact on the measured learning results are all key annotation attributes, and the key annotation attributes are screened out and the target learning data under the key annotation attributes are obtained, and multiple classification models are constructed and trained based on the target learning data, wherein the training of multiple classification models is based on the training data set obtained by dividing the multi-type learning data under the key annotation attributes of the second database.

[0076] In the specific implementation process of step S102, there is an embodiment in which: the step of obtaining target learning data with key annotated attributes from the learning data with annotated attributes based on the influence coefficient includes:

[0077] S1021, judging whether a preset relevance condition is met based on the calculated influence coefficient of the annotation attribute on the student's measurement learning result;

[0078] S1022: If yes, determine that the corresponding annotation attribute is a key annotation attribute, and the corresponding learning data is target learning data.

[0079] In steps S1021-S1022, the influence coefficient of the annotation attribute on the student's measured learning results is calculated by the preset calculation method Pearson correlation coefficient PCC, and it is determined whether the influence coefficient meets the preset correlation condition. If the preset correlation condition is whether the influence coefficient exceeds the preset threshold, the preset threshold can be specifically set according to the actual situation. If the preset correlation condition is met, the annotation attribute that meets the preset correlation condition is determined to be the key annotation attribute, wherein the influence coefficient is related to the correlation, that is, the higher the correlation between the annotation attribute and the measured learning result, the greater the influence coefficient. If the preset correlation condition is not met, the annotation attribute that does not meet the preset correlation condition is determined to be an ordinary annotation attribute, that is, the annotation attribute has a small impact on the measured learning result, and there is no need to construct the second database based on the learning data corresponding to the ordinary annotation attribute.

[0080] In step S103, after obtaining the target learning data of key labeled attributes and constructing a second database based on the target learning data of key labeled attributes, multiple classification models are constructed and trained based on the target learning data in the second database. The multiple classification models constructed at this time greatly improve the accuracy of the learning data to be tested. The multiple classification models include a random forest classification model RF, a support vector machine SVM, a logistic regression LR, a decision tree classification model DT, a gradient boosting machine GBM, a Gaussian naive Bayes classification model GNB and a multi-layer perception classification model MLP. The specific construction and training methods of some classification models are as follows.

[0081] Random Forest (RF) is an ensemble learning method that combines multiple decision trees to create a more robust model. This method is like its name, a forest or a group of trees. The tree-based classification model creates trees in stages and selects the best tree using a voting method. It also provides a relatively reliable indication of the usefulness of the features. This process consists of four steps:

[0082] 1. Randomly select a set of samples

[0083] 2. Create a tree and get prediction results

[0084] 3. Assign voting scores to each tree

[0085] 4. Select the tree with the most votes

[0086] In random forest, each decision tree makes a prediction based on the features of the student and the final prediction is based on the majority vote of the decision trees. It allows the algorithm to handle non-linear relationships between features and target classes, making it a popular choice for many classification problems.

[0087] Regression formula:

[0088]

[0089] Here, Formula 1 represents the regression equation for the random forest classification model. Where B is the number of data points, r f is the value returned by the model and b is the actual value of data point i.

[0090] Support Vector Machine (SVM) is a popular machine learning algorithm that can be used to solve multi-class classification problems, including evaluating student performance. SVM works by finding a hyperplane that separates the data into different categories with the maximum margin. The data points that are closest to the hyperplane are called support vectors, and they play a key role in determining the hyperplane. In the context of student performance evaluation, SVM is used to predict the class of students based on their attributes such as grades and attendance. The algorithm handles multi-class classification problems by creating multiple binary classification models and combining their results to come up with a final prediction. The results are then compared to other results, as well as one against another. The classification model with the highest accuracy score is considered the final result. The real benefit of SVM is that it works well on data that can be linearly separable, even without tuning.

[0091] The following is the formula for SVM:

[0092] y i (w·x i +b)≥1ζ i , i=1...mFormula 2

[0093] Formula 2 represents the decision boundary of the SVM model. In SVM, the goal is to find an optimal hyperplane that separates data points into different categories by maximizing the interval between categories. Where w is the weight vector, x is the input vector, and b is the bias.

[0094] The optimization problem of SVM is given by:

[0095]

[0096] Equation 3 presents the optimization problem or Lagrangian dual problem for SVM: instead of minimizing w and b under constraints involving a's, we can maximize a (the dual variable) under the previously obtained relationship between w and b.

[0097] Logistic regression (LR) is a statistical method used to solve classification problems. It is a generalized linear model used to model the relationship between a binary outcome variable and a set of input features. The basic principle of LR is to find the best linear combination of input features to maximize the probability of the observed outcome.

[0098] LR can be used to predict students’ proficiency on a particular learning object when evaluating student knowledge assessments. It can also be used to predict students’ performance on future assessments, thereby improving the teaching and learning process. LR is a simple, interpretable, and efficient model that can handle both linear and nonlinear relationships between student performance and knowledge and the features that influence them. In addition, LR can also be used for multi-class problems where the dataset contains more than two categories. It uses a logarithmic function to model the relationship between independent variables and binary dependent variables. In the context of student performance assessment, the independent variables may be attributes of the students, such as study habits, socioeconomic status, etc. In contrast, the dependent variables are the performance levels of the students, which are divided into multiple categories. LR estimates the probability of the response variable in each category and takes the category with the highest probability as the predicted category.

[0099] LR calculates the probability of an event (0 and 1) given a set of independent factors. Therefore, the data needs to be standardized before processing.

[0100] The logarithmic function used in the LR model is the softmax function, which converts the linear combination of independent variables into probability distributions for different categories. The softmax function is defined in Equation 4:

[0101]

[0102] Where P i is the predicted probability of category i, z i is a linear combination of the independent variables for category i, and k is the total number of categories. To calculate z for each category i , the LR model estimates the coefficients of the independent variables in a manner similar to the binary case. However, instead of estimating a single set of coefficients, LR estimates a separate set of coefficients for each category for multiple categories, resulting in a coefficient matrix.

[0103] For LR with multiple categories, a softmax function is used to model the probability of each category given a set of independent variables. The model estimates a coefficient matrix to calculate the linear combination of the independent variables for each category, and the data needs to be standardized to avoid overfitting.

[0104] The decision tree (DT) classification model is an algorithm for supervised learning tasks such as classification and regression. It works by recursively partitioning a dataset into subsets based on the values ​​of the input features. The goal is to create partitions that can lead to maximum separation between categories. The partitioning process produces a tree-like structure in which each internal node represents a feature test and each leaf node represents a class label. The basic principle of a decision tree is to learn to approximate any complex function by training on a set of input-output pairs. The training process is based on finding the best feature test at each internal node to achieve maximum separation between categories. The final decision tree is a set of feature tests, each of which represents a path from the root node of the tree to a leaf node. When evaluating student knowledge concept assessments, decision tree classification models classify students into different proficiency levels of a specific learning object, such as novice, beginner, intermediate, or expert. Decision trees are also used to predict student performance in future assessments, which can be used to improve the teaching and learning process. Decision trees can learn nonlinear relationships between student performance and knowledge and the features that affect them, and the results are simple and easy to understand. In addition, decision trees are also used to classify multi-class problems where the dataset contains more than two categories.

[0105] Decision tree classification model is a tree based classification model that uses two different types of nodes (decision nodes and leaf nodes). Decision nodes make decisions and have many branches, while leaf nodes represent the outcome of the decision and do not have any additional components. The decision process is based on the characteristics of the dataset. Attribute selection metrics are used to select the best features for the root node and child nodes. A question is asked in a decision tree and the answer (yes / no) determines which subtrees are included. Here, the starting decision node represents the root node followed by the subtrees, such as Figure 2 shown.

[0106] The Naive Bayes family of classification models, including the Gaussian Naive Bayes classification model (GNB), uses Bayes' theorem to classify data. GNB assumes that the continuous values ​​associated with each feature are spread out according to a Gaussian distribution. When looking for features that affect classification problems, Naive Bayes classification models can find them with very little data. GNB is a machine learning algorithm that can be used for multi-class problems where students' performance is evaluated. It is based on Bayes' theorem and assumes that features are conditionally independent given a class variable. The algorithm calculates the probability of each class given a feature and assigns the class with the highest probability as the final prediction. GNB is easy to implement and can handle both continuous and discrete features. However, it can perform poorly when features are highly correlated.

[0107] Gaussian Naive Bayes (GNB) supports the use of continuous-valued features in Equation 5, which also models each feature according to a Gaussian (normal) distribution.

[0108]

[0109] Where P(t) = π, parameter π t and μ are learned by maximum likelihood estimation. The above equation uses a Gaussian distribution and x i This dependency is encoded in the covariance matrix, where the letters in Formulas 1-5 are only used for illustration and have no actual meaning.

[0110] After obtaining multiple trained classification models, in order to ensure the advantages of the trained multiple classification models, the multiple classification models are tested and test results are generated. Based on the test results, evaluation is performed from multiple evaluation dimensions to obtain evaluation results corresponding to each classification model. According to the evaluation results of the multiple classification models, the target classification model is determined from the trained multiple classification models. The target classification model is the best classification model, which performs well in multiple evaluation dimensions. The multiple evaluation dimensions include classification accuracy, precision, recall rate, prediction error rate and area under the receiver operating characteristic curve, which are accuracy, precision, recall rate, MAE, MSE, RMSE, RAE and AUC respectively. The prediction error rate includes MAE, MSE, RMSE, RAE, and then the multiple classification models are accurately evaluated. Various evaluation dimensions are used to determine the confusion or prediction error of multiple classification models. A special condition table with two dimensions, which contains actual and expected results, and the same set of "categories", is used to describe confusion. The confusion matrix development helps to understand the performance indicators of each classification model performance, as shown in Figure 2.

[0111] Table 2 Confusion matrix

[0112]

[0113] In the specific process of step S103, there is an embodiment in which: the multiple classification models are evaluated from multiple evaluation dimensions to obtain multiple evaluation results, and a target classification model is determined based on the multiple evaluation results, including:

[0114] S1031, evaluating the multiple classification models respectively based on multiple evaluation dimensions to obtain an evaluation result corresponding to each classification model;

[0115] S1032: Select the best evaluation result from the multiple evaluation results, and determine the classification model corresponding to the best evaluation result as the target classification model.

[0116] In steps S1031-S1032, the corresponding calculation method is used based on the evaluation dimension to calculate the multiple classification models respectively, and the evaluation results corresponding to each classification model are obtained. Based on the evaluation results, the best evaluation result is selected according to the preset selection method, and the classification model corresponding to the best evaluation result is determined as the target classification model, wherein the preset selection method can be to select one of the evaluation dimensions with excellent performance, or to assign different weights to the evaluation results, so as to determine the target classification model based on the evaluation results. As shown in Table 3, these are the evaluation results corresponding to multiple classification models.

[0117] According to Table 3, GBM (Gradient Boosting Machine) exhibits the highest prediction accuracy of 98%. In terms of prediction error, GBM also performs well. In terms of performance, RF (Random Forest), DT (Decision Tree), and MLP (Multi-layer Perceptron) are comparable to GBM. The performance may be improved when using linear classification models such as SVM (Support Vector Machine) or LR (Logistic Regression). At this test level, the tree-based classification model performs better;

[0118] The tree-based classification models outperformed the linear classification models in terms of precision and recall. GBM (gradient boosting machine) achieved 99% prediction accuracy, DT (decision tree) and RF (random forest) achieved 98% prediction accuracy. GBM demonstrated 97% accurate prediction recall. On the dataset, MLP (multi-layer perceptron) achieved 96% accuracy, 95% precision, and 90% recall, performing well in multi-classification. When considering prediction error, GBM performed best among other classification models;

[0119] Regarding error metrics, the lowest MAE (mean absolute error), MSE (mean square error), and RMSE (root mean square error) for the DT (decision tree) and GBM (gradient boosting) classification models were 0.03, 0.03, and 0.1, respectively, for both DT and GBM. The highest scores were found in the LR (logistic regression) classification model, which were 0.09, 0.09, and 0.3, respectively. The RAE (root absolute error) scores ranged from 0.051 (DT and GBM) to 0.12 (LR). The highest AUC (area under the receiver operating characteristic curve) score was for the GBM classification model (0.97), and the lowest was for the SVM (support vector machine) classification model (0.92).

[0120] Table 3 Evaluation results of multiple classification models

[0121]

[0122] Therefore, the GBM (Gradient Boosting Machine) classification model has the highest prediction accuracy of 98% and performs well in terms of prediction error; Two experiments were also conducted to understand the impact of reducing feature vectors on the prediction accuracy of the classification model, so GBM (Gradient Boosting Machine) is the target classification model.

[0123] In the specific process of step S1031, there is an embodiment in which the multiple classification models are evaluated respectively based on multiple evaluation dimensions to obtain the evaluation result corresponding to each classification model, including:

[0124] S10311. Determine multiple evaluation dimensions and test data sets for multiple trained classification models;

[0125] S10312. Based on the test data set, the multiple classification models are tested from multiple evaluation dimensions to obtain corresponding test results, and based on the test results, the multiple classification models are evaluated to obtain corresponding evaluation results.

[0126] In steps S10211-S10312, after the multiple classification models are trained, multiple evaluation dimensions and test data sets for the trained multiple classification models are determined based on the characteristics of the multiple classification models, and the trained multiple classification models are tested using the test data sets to generate test results. The multiple evaluation dimensions can better reflect the differences between each of the multiple models, and calculations are performed based on the multiple evaluation dimensions based on the test results to obtain evaluation results corresponding to each evaluation dimension.

[0127] The test data set is obtained by dividing the learning data under the key annotation attributes stored in the second database, and the second database is also divided into training data sets, such as Figure 3 Shown is a schematic diagram of the relationship between the second database and the multiple classification models.

[0128] Common strategies for using the second database to divide into training data sets and test data sets include: randomly dividing the data set into training sets and test sets. Usually, the training set accounts for 70%-80% and the test set accounts for 20%-30%. This division method is simple, but may introduce bias, especially when the data distribution is uneven; in classification tasks, in order to ensure that the proportion of each category in the training data set and the test data set is the same as the proportion in the overall data set, a stratified sampling method can be used. This method is particularly suitable for data sets with unbalanced categories; in time series data or data with time dependence, early data is usually used as a training set and later data is used as a test data set to simulate prediction scenarios in the real world. If the second database is a data set containing 1,000 samples, each of which has five data attributes PEG, LPR, STR, SCG, STG) and a measurement learning outcome UNS, indicating the student's knowledge level, may be a categorical variable such as "very low", "low", "medium", "high".

[0129] Example 1: Random Partition: Training Dataset: 700 samples (randomly selected); Testing Dataset: 300 samples (remaining samples) In this case, the dataset is randomly divided into training and testing datasets, ensuring that both have statistically similar distributions.

[0130] Example 2: Stratified sampling: Assume that the target variable (UNS) has four categories: "very low", "low", "medium", and "high", with 250 samples in each category; training dataset: 175 samples (70%) are randomly selected from each category, for a total of 700 samples; test dataset: the remaining 75 samples (30%) in each category, for a total of 300 samples. In this case, the proportion of each category in the training dataset and the test dataset is the same as that in the overall dataset, ensuring that the performance of the model on different categories can be evaluated.

[0131] Before the training data set and the test data set are used, data cleaning (removing missing values, outliers, etc.), feature scaling (standardization or normalization), feature encoding (such as One-Hot encoding), etc. are required, and 5-fold cross validation is used to evaluate the performance of multiple classification models to reduce the risk of overfitting.

[0132] Where precision is called the metric used to determine the accuracy of accurate predictions (Formula 7); recall is called the ratio of accurate predictions divided by all estimates (Formula 8); recall is a valuable metric when false negatives exceed false positives; accuracy is called the ratio of all correct predictions to all correct and incorrect predictions (Formula 6).

[0133]

[0134] Recall rate = TP (TP + FN) Formula 8

[0135]

[0136]

[0137] Among them, the test size is n, y i is the predicted value, The average of the actual values. The RMSE, MAE, MES, and RAE in the prediction error rate are all existing formulas, and the meanings of the letters are consistent with those in the prior art. In addition to these measurements, the AUC (area of ​​the receiver operating characteristic curve) is also calculated, that is, the ratio of correctly classified positive samples (TPR) to misclassified negative samples (FPR), and plotted at all potential thresholds to form a graph called the receiver operating characteristic (ROC) curve.

[0138] In step S104, the learning data to be tested generated by the student to be tested for the second digital courseware is obtained. The second digital courseware and the first digital courseware are both in the digital courseware library of the digital platform. The second digital courseware and the first digital courseware may correspond to the same or different ones. As shown in step S101, the corresponding annotation attributes are determined according to the data attributes of the learning data to be tested, and the data attributes are stored under the corresponding annotation attributes. At this time, the learning data to be tested with the annotation attributes are obtained, and the learning data to be tested with the annotation attributes are input into the target classification model to output the classification results. The classification standard is to divide the students' knowledge levels according to their personal behaviors and performance data in the digital courseware learning platform. Therefore, based on the classification results, the learning results of the students to be tested for the knowledge content in the second digital courseware are measured and the knowledge level of the students to be tested is mastered.

[0139] This application divides the students to be tested into four categories according to their knowledge level, and the classification results are "very low", "low", "medium" and "high", which correspond to the students' proficiency in specific learning objects, such as novice, beginner, intermediate or expert. Very low: indicates that the student's mastery of the learning object is very low. They may have just started learning and are not very familiar with the relevant knowledge. Low: indicates that the student's mastery of the learning object is low. They may have mastered some basic knowledge, but still need further study and practice. Medium: indicates that the student's mastery of the learning object is medium. They have mastered certain knowledge, but there is still room for improvement. High: indicates that the student has a high mastery of the learning object, has mastered the relevant knowledge proficiently, and can apply this knowledge to solve complex problems. Therefore, the digital courseware updates the courseware for students based on the classification results, so that students can learn courseware that meets their knowledge level.

[0140] This application aims to develop a machine learning-based system that can accurately assess students' performance and knowledge level throughout the process of student learning. By analyzing a large amount of student data and identifying the key annotation attributes that have the greatest impact on performance, educators and institutions can gain insights on how to better support students. One benefit of this application is that it allows educators to adjust their teaching and resources according to the specific needs of each student. By identifying areas where students need additional support or challenges, educators can provide more targeted interventions to help students succeed, which helps reduce the waste of resources and optimize their use, thereby promoting a more sustainable education system. In addition, by using machine learning technology, this application demonstrates the effects of educational technology that can further reduce the environmental impact of traditional teaching methods. For example, digital courseware can reduce the need for physical classrooms and textbooks, which can lead to resource conservation and reduced waste. In general, the present invention presents how to use machine learning technology to help organizations and individuals improve resource utilization in a more optimized way to maintain sustainability, by adjusting teaching and resources according to the specific needs of each student, and using more sustainable educational technology.

[0141] Example 2

[0142] The present application also provides a device for measuring the learning results of knowledge content in digital intelligent courseware, such as Figure 4 The block diagram of a device for measuring the learning results of knowledge content in digital intelligent courseware is shown. The function implemented by the device for measuring the learning results of knowledge content in digital intelligent courseware corresponds to the steps of executing a method for measuring the learning results of knowledge content in digital intelligent courseware on a terminal device. The device can be understood as a component of a server including a processor, and the device includes:

[0143] The configuration module 401 is used to configure the annotation attributes for measuring the learning results of the first digital courseware, and obtain multiple types of learning data under the annotation attributes generated by students learning the first digital courseware within a preset historical time period;

[0144] A calculation module 402 is used to calculate the influence coefficient of the annotation attribute corresponding to the multi-type learning data on the measurement learning result, and obtain the target learning data under the key annotation attribute from the multi-type learning data based on the influence coefficient;

[0145] A training module 403 is used to construct and train multiple classification models based on the target learning data, and determine a target classification model from the trained multiple classification models;

[0146] The measurement module 404 is used to obtain the learning data to be tested of the student to be tested for the second digital courseware, and input the learning data to be tested into the target classification model to output the classification result, and measure the learning result of the student to be tested for the knowledge content in the second digital courseware based on the classification result.

[0147] In a feasible implementation manner, the configuration module includes:

[0148] A generating module, configured to generate a plurality of attribute fields based on pre-configured annotation attributes, and to acquire and store corresponding learning data according to the attribute fields to construct a first database;

[0149] A first acquisition module, used to acquire, from a first database, multiple types of learning data generated by students for a first digital courseware;

[0150] In a feasible implementation, the configuration module also includes:

[0151] A second acquisition module, used to acquire target learning data under key annotation attributes from the first database;

[0152] A construction module is used to construct a second database based on the target learning data, so as to construct and train multiple classification models based on the target learning data in the second database.

[0153] In a feasible implementation manner, the configuration module further includes:

[0154] A first determination module, used to determine a plurality of attribute fields corresponding to the annotation attribute;

[0155] The matching module is used to match multiple data attributes corresponding to the attribute field to obtain multiple types of learning data matching the attribute field.

[0156] In a feasible implementation manner, the configuration module further includes:

[0157] A third acquisition module is used to acquire a mapping relationship between the annotation attribute and the attribute field;

[0158] The second determination module is used to determine a plurality of attribute fields corresponding to the annotation attribute according to a preset mapping relationship.

[0159] In a feasible implementation manner, the computing module includes:

[0160] A first calculation module is used to calculate the influence coefficient of the annotation attribute on the student's measured learning results based on the calculation network model, and determine whether the influence coefficient meets the preset correlation condition;

[0161] The third determination module is used to determine that if yes, the corresponding annotation attribute is a key annotation attribute, and the corresponding learning data is target learning data.

[0162] In a feasible implementation manner, the measurement module includes:

[0163] A second calculation module is used to evaluate the multiple classification models respectively based on multiple evaluation dimensions to obtain an evaluation result corresponding to each classification model;

[0164] The selection module is used to select the best evaluation result from the multiple evaluation results, and determine the classification model corresponding to the best evaluation result as the target classification model.

[0165] In a feasible implementation manner, the measurement module further includes:

[0166] A fourth determination module is used to determine multiple evaluation dimensions and test data sets for multiple trained classification models;

[0167] An evaluation module is used to test the multiple classification models from multiple evaluation dimensions based on the test data set to obtain corresponding test results, and to evaluate them based on the test results to obtain corresponding evaluation results.

[0168] Example 3

[0169] The present application also provides an electronic device, such as Figure 5 As shown, it includes: a processor 501, a memory 502 and a bus 503, the memory 502 stores machine-readable instructions executable by the processor 501, when the electronic device is running, the processor 501 communicates with the memory 502 through the bus 503, and when the machine-readable instructions are executed by the processor 501, any one of the steps of a method for measuring the learning results of knowledge content in digital intelligent courseware is performed.

[0170] Example 3

[0171] The present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the computer program executes the steps of any one of the methods for measuring the learning results of knowledge content in digital intelligent courseware.

[0172] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the system and device described above can refer to the corresponding process in the method embodiment, and will not be repeated in this application. In the several embodiments provided in this application, it should be understood that the disclosed system, device and method can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation. For example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0173] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0174] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0175] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a platform server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard drives, ROM, RAM, magnetic disks, or optical disks.

[0176] The above are only specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for measuring the learning results of knowledge content in digital intelligent courseware, characterized in that: The method comprises: Configuring a labeling attribute for measuring a learning result of the first digital courseware, and obtaining multiple types of learning data under the labeling attribute generated by students learning the first digital courseware within a preset historical time period; Calculating the influence coefficient of the annotation attribute corresponding to the multi-type learning data on the measurement learning result, and obtaining the target learning data under the key annotation attribute from the multi-type learning data based on the influence coefficient; Constructing and training multiple classification models based on the target learning data, and determining a target classification model from the trained multiple classification models; Acquire the learning data of the student to be tested for the second digital courseware, input the learning data to be tested into the target classification model to output the classification result, and measure the learning result of the student to be tested for the knowledge content in the second digital courseware based on the classification result.

2. The method according to claim 1, characterized in that The obtaining of the multi-type learning data under the annotated attributes generated by the student learning the first digital courseware within a preset historical time period includes: Generate a plurality of attribute fields based on pre-configured annotation attributes, and acquire and store corresponding learning data according to the attribute fields to construct a first database; Acquire, from the first database, multiple types of learning data generated by students for the first digital courseware; The obtaining of target learning data under key annotation attributes includes: Acquire target learning data under key annotation attributes from the first database; A second database is constructed based on the target learning data, so as to construct and train multiple classification models based on the target learning data in the second database.

3. The method according to claim 1, characterized in that The obtaining of the multi-type learning data under the annotated attributes generated by the student learning the first digital courseware within a preset historical time period includes: Determine a plurality of attribute fields corresponding to the annotation attribute; Matching is performed according to multiple data attributes corresponding to the attribute field to obtain multiple types of learning data matching the attribute field.

4. The method according to claim 3, characterized in that The determining of the multiple attribute fields corresponding to the annotation attribute includes: Obtaining a mapping relationship between the annotation attribute and the attribute field; According to the mapping relationship, a plurality of attribute fields corresponding to the annotation attribute are determined.

5. The method according to claim 1, characterized in that The acquiring target learning data under key annotation attributes from the multi-type learning data based on the influence coefficient includes: Calculating the influence coefficient of the annotation attribute on the student's measured learning results based on the computing network model, and judging whether the influence coefficient meets the preset correlation condition; If so, the corresponding annotation attribute is determined to be the key annotation attribute, and the corresponding learning data is determined to be the target learning data.

6. The method according to claim 1, characterized in that The step of determining a target classification model from a plurality of trained classification models comprises: Evaluate the multiple classification models respectively based on multiple evaluation dimensions to obtain evaluation results corresponding to each classification model; The best evaluation result is selected from the multiple evaluation results, and the classification model corresponding to the best evaluation result is determined as the target classification model.

7. The method according to claim 6, characterized in that The multiple classification models are evaluated respectively based on multiple evaluation dimensions to obtain evaluation results corresponding to each classification model, including: Determine multiple evaluation dimensions and test data sets for multiple trained classification models; Based on the test data set, the multiple classification models are tested from multiple evaluation dimensions to obtain corresponding test results, and based on the test results, they are evaluated to obtain corresponding evaluation results.

8. A device for measuring the learning results of knowledge content in digital intelligent courseware, characterized in that: The device comprises: A configuration module, used to configure a labeling attribute for measuring a learning result of the first digital courseware, and to obtain multiple types of learning data under the labeling attribute generated by students learning the first digital courseware within a preset historical time period; A calculation module, used to calculate the influence coefficient of the annotation attribute corresponding to the multi-type learning data on the measurement learning result, and obtain the target learning data under the key annotation attribute from the multi-type learning data based on the influence coefficient; A training module, used to construct and train multiple classification models based on the target learning data, and determine a target classification model from the trained multiple classification models; The measurement module is used to obtain the learning data to be tested of the students for the second digital courseware, input the learning data to be tested into the target classification model to output the classification result, and measure the learning results of the students to be tested for the knowledge content in the second digital courseware based on the classification result.

9. An electronic device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor and the memory communicate via the bus. When the machine-readable instructions are executed by the processor, the steps of a method for measuring learning outcomes of knowledge content in digital intelligent courseware are performed as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, executes the steps of a method for measuring learning results of knowledge content in digital intelligent courseware as described in any one of claims 1 to 7.

Images