Multi-dimensional education data analysis and decision support system
By analyzing the learning status labels and sequences of target objects, identifying problems in curriculum and teaching methods, the problem of unreliable decision-making in the existing technology is solved and more accurate educational support is achieved.
Patent Information
- Application Number
- CN202510487497.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-18
- Publication Date
- 2025-07-25
AI Technical Summary
The existing technology is difficult to identify problems in curriculum and teaching methods and cannot provide effective decision-making advice for educational platforms.
By analyzing the performance of the target object in course learning, setting learning status tags, establishing learning status sequences, comparing learning status sequences in the same class and different classes, identifying individual, class or course problems, and improving decision-making reliability.
Improve the reliability of educational decision-making, can identify individual, class or course problems, and provide targeted support measures.
Smart Images
Figure CN120373653A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of educational decision-making support, and specifically relates to a multi-dimensional educational data analysis and decision-making support system. The present invention belongs to the field of educational decision-making support, and specifically relates to a multi-dimensional educational data analysis and decision-making support system. Background Art
[0002] With the wide application of big data technology in the field of education, educational methods are becoming increasingly diverse. In educational activities, big data technology can realize personalized learning path recommendation, teaching method optimization, etc., and can also be used to monitor students' learning behaviors in real time to achieve precise management.
[0003] In the process of using big data technology to monitor students for precise management in existing solutions, machine learning algorithms are mainly used to analyze students' learning behaviors and learning outcomes, identify their mastery of knowledge points in the course, and then generate targeted reinforcement strategies. Although this solution can improve students' mastery of knowledge points through reinforcement, it only focuses on students' learning effects, is difficult to identify problems existing in the course or even teaching methods, and cannot provide effective decision-making suggestions for educational platforms.
[0004] The present invention provides a multi-dimensional educational data analysis and decision-making support system to solve the above technical problems. Summary of the Invention
[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; for this purpose, the present invention proposes a multi-dimensional educational data analysis and decision-making support system, which sets learning status labels by analyzing the performance of target objects in course learning to provide a data basis for analyzing whether there are personal problems; then establishes a learning status sequence with the learning status labels, judges whether there are personal problems by comparing the learning status sequences of target objects in the same class, and judges whether there are class problems or course problems by comparing the learning status sequences of target objects in different classes. Identifying problems through multi-dimensional educational data can improve the reliability of relevant decisions.
[0006] To achieve the above object, the first aspect of the present invention provides a multi-dimensional educational data analysis and decision-making support system, including an analysis and decision module and a data processing module connected thereto; Data processing module: used to collect the behavior data and learning outcomes of a number of target objects associated with the target course; wherein, a number of target objects associated with the target course are the applicable objects of the course; and, used to analyze the behavior data to obtain the concentration label of the target object, and analyze the learning outcomes to obtain the knowledge mastery sequence of the target object; wherein, the learning outcomes include exam scores and interaction scores; Analysis and Decision-making Module: used to identify the knowledge distribution sequence in the target course. After time-aligning the knowledge distribution sequence, attention label, and knowledge mastery sequence, calculate the learning status label of the target object based on the attention label and knowledge mastery sequence; and, used to construct the learning status sequence of the target object according to the learning status label; jointly judge the learning status sequences of several target objects to determine whether there are preset problems; if so, formulate support measures based on the existing preset problems; where the preset problems include individual problems, class problems, or course problems.
[0007] Preferably, analyzing the behavioral data to obtain the attention label of the target object includes: Extracting the behavioral data of the target object and extracting the attention features from the behavioral data; where the behavioral data includes video data, and the attention features include facial features, eye movement features, and behavioral features; Judging the attention label corresponding to the attention features through a machine learning algorithm; where the attention label corresponds to high attention, medium attention, or low attention.
[0008] Preferably, parsing the learning achievements to obtain the knowledge mastery sequence of the target object includes: Identifying several knowledge points examined from the learning achievements; Identifying the mastery situation of the target object for several knowledge points in the target course from the learning achievements, setting mastery labels according to the mastery situation, and associating the mastery labels with their corresponding knowledge points.
[0009] Preferably, identifying the knowledge distribution sequence in the target course includes: Extracting the target course, identifying and marking several knowledge points in the target course; Matching the associated moments of several knowledge points in the target course, arranging several knowledge points in the order of the associated moments, and obtaining the knowledge distribution sequence.
[0010] Preferably, time-aligning the knowledge distribution sequence, attention label, and knowledge mastery sequence includes: Extracting several knowledge points in the knowledge distribution sequence and integrating them into a knowledge point group; Matching several knowledge points in the knowledge mastery sequence in the knowledge point group, and associating the mastery label corresponding to the successfully matched knowledge point with its associated moment; Matching the attention label through the associated moments of several knowledge points in the knowledge point group, and associating the successfully matched attention label with the associated moment.
[0011] Preferably, calculating the learning status label of the target object based on the attention label and knowledge mastery sequence includes: Extracting the mastery label and attention label associated with the associated moment; Match the learning status tags corresponding to the mastery tag and the concentration tag from the preset tag group; wherein, the preset tag group includes the learning status tags corresponding to different combinations of the mastery tag and the concentration tag.
[0012] Preferably, construct a learning status sequence of the target object according to the learning status tags, including: Arrange the associated moments of several knowledge points in the target course in chronological order to generate a time sequence; wherein, the associated moment is the occurrence moment of the knowledge point in the target course; Extract the learning status tags of the target object; use the associated moments in the time sequence as independent variables and the learning status tags as dependent variables to construct the learning status sequence of the target object.
[0013] Preferably, jointly judge the learning status sequences of several target objects, including: Integrate the learning status sequences of several target objects into several curve combinations according to the preset questions; wherein, the curve combination includes Curve Group 1 and Curve Group 2. Curve Group 1 is used to identify individual problems, and Curve Group 2 is used to identify class problems or course problems; Analyze several curve combinations in the problem recognition order through a machine learning algorithm to determine whether there are preset problems; wherein, the problem recognition order is individual problems, class problems, and course problems in sequence.
[0014] Preferably, integrate the learning status sequences of several target objects into several curve combinations according to the preset questions, including: When the preset problem is an individual problem, integrate the learning status sequence of the target object according to the set class to obtain Curve Group 1; When the preset problem is a class problem or a course problem, integrate the learning status sequences of the target objects in different classes to obtain Curve Group 2.
[0015] Preferably, analyze several curve groups in the problem recognition order through a machine learning algorithm, including: Classify Curve Group 1 through a machine learning algorithm to determine whether there are individual problems; if so, extract the target objects with individual problems and the corresponding knowledge points; if not, proceed to the next step; Classify Curve Group 2 through a machine learning algorithm to determine whether there are class problems; if so, extract the class with class problems and the corresponding knowledge points; if not, determine that there are course problems.
[0016] Compared with the prior art, the beneficial effects of the present invention are: The present invention analyzes behavioral data to obtain the concentration labels of the target object, and analyzes the learning outcomes to obtain the knowledge mastery sequence of the target object; identifies the knowledge distribution sequence in the target course, and after time-aligning the knowledge distribution sequence, concentration labels, and knowledge mastery sequence, calculates the learning status labels of the target object according to the concentration labels and knowledge mastery sequence; constructs the learning status sequence of the target object based on the learning status labels; jointly judges the learning status sequences of several target objects to determine whether there are preset problems; by analyzing the performance of the target object in course learning to set learning status labels, the present invention provides a data basis for analyzing whether there are personal problems; then establishes a learning status sequence with the learning status labels, and determines whether there are personal problems by comparing the learning status sequences of target objects in the same class, and determines whether it is a class problem or a course problem by comparing the learning status sequences of target objects in different classes. Identifying problems through multi-dimensional educational data can improve the reliability of relevant decisions. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.
[0018] Figure 1 It is a schematic diagram of the method steps of the multi-dimensional educational data analysis and decision-making system of the present invention; Figure 2 It is a schematic diagram of the system principle of the multi-dimensional educational data analysis and decision-making system of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0019] The following will clearly and completely describe the technical solutions of the present invention in conjunction with the embodiments. Obviously, the described embodiments are only some of the embodiments of the present invention, rather than all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.
[0020] Existing analysis solutions based on multi-dimensional education data either judge whether there are problems with students themselves by analyzing students' behaviors during classes, or judge whether there are problems with teachers' teaching by analyzing teachers' performances during classes. However, simply analyzing students or teachers through behavioral data lacks reliability. That is, it is difficult to identify specific problems that lead to a student's learning effect not meeting requirements through the behavioral data of a certain student individual. Individual problems, teachers' teaching problems, and curriculum setting problems may all affect their learning effects. Therefore, the existing solutions cannot provide reliable support for educational decision-making. The technical solution of the present invention can solve the above technical problems, and the technical solution will be described in detail below. Embodiment 1:
[0021] Please refer to Figure 1 - Figure 2 , an embodiment of the first aspect of the present invention provides a multi-dimensional education data analysis and decision support system, including an analysis and decision module, and a data processing module connected thereto; Data processing module: used to collect the behavioral data and learning achievements of several target objects associated with the target course; and, used to analyze the behavioral data to obtain the concentration labels of the target objects, and parse the learning achievements to obtain the knowledge mastery sequences of the target objects; Analysis and decision module: used to identify the knowledge distribution sequence in the target course, after time-aligning the knowledge distribution sequence, concentration labels, and knowledge mastery sequences, calculate the learning status labels of the target objects according to the concentration labels and knowledge mastery sequences; and, used to construct the learning status sequences of the target objects according to the learning status labels; jointly judge the learning status sequences of several target objects to determine whether there are preset problems; if so, formulate support measures according to the existing preset problems.
[0022] The main idea of the present invention is as follows: First, use students as target objects to collect and process multi-dimensional data, respectively identify the concentration and learning effects of the target objects, and evaluate the mastery of each knowledge point by the target objects; in the case where the mastery of knowledge points does not meet the requirements, if it is the problem of the target object itself, the insufficient knowledge points can be strengthened, and if it is not the problem of the target object itself, then jointly judge whether there are class problems and curriculum problems among multiple target objects. Class problems mainly include teachers' teaching problems or class atmosphere problems.
[0023] In order to reduce the data processing volume, still start from the target object (student). First, collect the behavioral data and learning achievements of several target objects associated with the target course; set the concentration labels of the target objects according to the behavioral data, and judge the mastery of each knowledge point by the target objects according to the learning achievements.
[0024] The target object refers to the students associated with the target course. The "association" in the present invention can be understood as that the students have selected the target course, or the platform (including the school) requires the students to select the target course. Behavioral data and learning outcomes are part of the multi-dimensional educational data. To improve the recognition accuracy of machine learning algorithms, other relevant educational data can be introduced; behavioral data is mainly used to identify the concentration of the target object and determine whether the target object focuses on learning during the learning process. Therefore, the behavioral data can include video data, and data collected by smart wearable devices can also be included if conditions permit; learning outcomes are mainly used to identify whether the target object has mastered the knowledge points, including exam scores or interactive scores in the course.
[0025] After analyzing and obtaining the concentration labels and knowledge mastery sequences of each target object, analyze the learning status labels of each target object at each knowledge point in the order of the knowledge points in the target course. The learning status label is jointly set through the concentration label and the knowledge mastery sequence.
[0026] The preset problems mainly include individual problems, class problems or course problems. Individual problems refer to the poor learning effect caused by the students' own problems. Class problems refer to the poor learning effect of students caused by the class atmosphere or teaching methods (instructors). Course problems refer to the mismatch between the setting of the target course and the needs or basic requirements of the target object, resulting in poor learning effects of students.
[0027] After obtaining the learning status labels of the target objects associated with the target course, arrange the learning status labels according to the knowledge point sequence in the target course to obtain the learning status sequence of the target object. By identifying the learning status sequences of the target objects in the same class, it can be identified whether there are problems with the target object itself; by identifying the learning status sequences of the target objects in different classes under the target course, it can be identified whether there are class problems or course problems.
[0028] Next, the key steps in the technical solution of the present invention will be described in detail.
[0029] As described above, in order to reduce the data processing volume, the present invention starts from the target object. Therefore, first collect the behavioral data and learning outcomes of several target objects associated with the target course. By extracting the subject records of the students from the database or the course records configured for the students by the platform, the students associated with each course can be obtained. Taking the course to be analyzed as the target course, the students associated with the target course are the target objects.
[0030] During the class, the class videos of the target objects corresponding to the target course are collected through the camera to obtain the video data of each target object. The video processing algorithm is used to identify whether there is interaction in the video data. If there is interaction, the interaction score is identified, and both the video data and the interaction score are associated with the corresponding target object. Of course, the facial features, eye features, and behavior features of the target object can be identified from the video data. The facial features, eye features, and behavior features are all used to identify the concentration of the target object, and they assist each other to improve the identification accuracy. During or after the target course, an exam paper will be set for the knowledge points in the target course, and the exam score of the target object will be associated with the target object and stored in the database.
[0031] Extract the behavior data of the target object from the database, and analyze the behavior data to obtain the concentration label of the target object, including: extracting the concentration features from the behavior data; judging the concentration label corresponding to the concentration features through the machine learning algorithm.
[0032] Exemplarily, preprocess the video data, such as denoising, grayscale transformation, and frame extraction, etc. Use the OpenFace toolbox to extract facial features, such as head pose, gaze direction, etc. The eye features mainly include features such as gaze direction, gaze dwell time, blink rate, and saccade. The behavior features mainly include listening attentively, taking notes, being distracted, etc. Integrate the facial features, eye features, and behavior features to form a multi-modal feature vector, and input the multi-modal feature vector into the trained support vector machine model to obtain the corresponding concentration label. The concentration label corresponds to high concentration, medium concentration, or low concentration The training of the support vector machine is mainly achieved through standard training data. The acquisition of the standard training data can extract the class videos of students from historical data (or a third-party database), extract the facial features, eye features, and behavior features of the students in the class videos, and then set the concentration label for these feature combinations to obtain sufficient standard training data.
[0033] Analyze the learning achievements of the target object to obtain the knowledge mastery sequence of the target object, including: identifying several knowledge points examined from the learning achievements; identifying the mastery situation of the target object for several knowledge points in the target course from the learning achievements, setting the mastery label according to the mastery situation, and associating the mastery label with its corresponding knowledge point.
[0034] Exemplarily, extract the knowledge points examined from the learning score or interaction score of the target object. If the target object answers the questions corresponding to the knowledge points correctly, it is determined that the target object has mastered the knowledge point. If the answer is wrong, it is determined that the target object has not mastered the knowledge point; when the target object masters the knowledge point, the mastery label can be set to 1, otherwise the mastery label can be set to 0. Finally, associate the set mastery label with the knowledge point corresponding to the target object.
[0035] The target course includes several knowledge points. In order to match them with the mastery situation of the target object, it is necessary to first clarify the knowledge points and their distribution in the target course. Specifically, it is to identify the knowledge distribution sequence in the target course, including: extracting the target course, identifying and marking several knowledge points in the target course; matching the associated moments of several knowledge points in the target course, and arranging several knowledge points in the order of the sequence of the associated moments to obtain the knowledge distribution sequence.
[0036] The knowledge points in the target course can be identified by machine learning algorithms for video data, or the PPT of the target course can be identified by generative large models. Of course, the knowledge points in the target course can also be marked by the teaching teacher or corresponding experts. Determine the occurrence moments of several knowledge points in the target course, use this occurrence moment as the associated moment of the corresponding knowledge point, and then arrange the knowledge points in the target course in the order of the sequence of the associated moments to obtain the knowledge distribution sequence of the target course.
[0037] The relevance among the foregoing attention tags, knowledge mastery sequence, and knowledge distribution sequence is not clear. First, align the three in terms of time, specifically including: Extracting several knowledge points in the knowledge distribution sequence and integrating them into a knowledge point group; matching several knowledge points in the knowledge mastery sequence in the knowledge point group, and associating the mastery tag corresponding to the successfully matched knowledge point with its associated moment; matching the attention tags through the associated moments of several knowledge points in the knowledge point group, and associating the successfully matched attention tags with the associated moments.
[0038] The knowledge mastery sequence records the mastery situation of the target object for several knowledge points. First, use these knowledge points to match several knowledge points in the knowledge distribution sequence. If the match is successful, it means that the knowledge point in the target course has been examined, and associate the mastery situation of this knowledge point with the associated moment of the corresponding knowledge point in the knowledge point group; if the match is not successful, it means that the corresponding knowledge point has not been examined and can be ignored. Similarly, the attention tags run through the entire target course, and match the attention tags corresponding to the associated moments of several knowledge points in the knowledge point group for the target object. In this way, for each examined knowledge point in the target course, the target object corresponds to an attention tag and a mastery situation at its associated moment.
[0039] The present invention determines the learning situation of the target object for knowledge points from two perspectives of attention and mastery. First, extract the preset tag group from the database. The preset tag group includes learning status tags corresponding to different combinations of mastery tags and attention tags, and match the learning status tags corresponding to the target object at each associated moment from the preset tag group.
[0040] Exemplarily, assume that the mastery label being 1 indicates that the target object has mastered the corresponding knowledge point, and 0 indicates that the corresponding knowledge point has not been mastered; the concentration label being 3 indicates high concentration of the target object when learning the corresponding knowledge point during the teaching process, 2 indicates medium concentration, and 1 indicates low concentration.
[0041] Then: 1) If the mastery label is 1 and the concentration label is 1, set the learning status label to 11; 2) If the mastery label is 0 and the concentration label is 1, set the learning status label to 01; 3) If the mastery label is 1 and the concentration label is 2, set the learning status label to 12; 4) If the mastery label is 0 and the concentration label is 2, set the learning status label to 02; 5) If the mastery label is 1 and the concentration label is 3, set the learning status label to 13; 6) If the mastery label is 0 and the concentration label is 3, set the learning status label to 03; The preset label group stored in the database includes the above six cases.
[0042] In order to be able to input the learning status labels of each target object into the machine learning algorithm for classification, the present invention constructs a learning status sequence of the target object according to the learning status label, including: Arrange the associated moments of several knowledge points in the target course in chronological order to generate a time series; extract the learning status label of the target object; use the associated moments in the time series as independent variables and the learning status label as the dependent variable to construct the learning status sequence of the target object.
[0043] After the above data is prepared, it can be determined whether there are preset problems through the machine learning algorithm. The judgment logic in this judgment process of the present invention can be summarized as follows: If the target object is not focused enough when learning a certain knowledge point and has not mastered the corresponding knowledge point, while the target objects in the same class (with the same teaching teacher, and the teaching location can be not considered) have all mastered (or most have mastered) the corresponding knowledge point under the condition of being focused, then it can be determined that it is a personal problem of the target object. If even when the target object studies attentively in the same class, most students have not mastered the corresponding knowledge point, then it may be a problem of the teaching teacher, or it may also be that the course itself is not set up properly. At this time, it is necessary to analyze by combining the learning status sequences of the target objects in other classes. If in multiple classes, only individual classes do not meet the requirements in terms of the mastery of a certain knowledge point, then it can be considered that there is a problem with the teaching method of the teaching teacher in that class; if the mastery of a certain knowledge point by the target objects in multiple classes does not meet the requirements, then it can be considered that there is a problem with the setting of this knowledge point in the target course.
[0044] Specifically, according to the preset questions, the learning status sequences of several target objects are integrated into several curve combinations, including: When the preset question is an individual question, the learning status sequences of the target objects are integrated according to the set classes to obtain the first curve group; when the preset question is a class question or a course question, the learning status sequences of the target objects in different classes are integrated to obtain the second curve group.
[0045] It can be understood that: according to the first curve group, it can be identified whether there is a problem with a certain target object in the corresponding class; if it is judged that it is not a personal problem of a certain target object, then it is necessary to combine the learning status sequences of the target objects in different classes to judge whether it is a problem of the teaching teacher in an individual class (that is, the teaching method problem). If it is not a problem of the teaching teacher either, then it can be determined as a curriculum setting problem.
[0046] After obtaining the curve combinations, several curve groups are analyzed according to the problem recognition order through machine learning algorithms, including: classifying the first curve group through machine learning algorithms to judge whether there are personal problems; if so, extracting the target objects with personal problems and the corresponding knowledge points; if not, proceeding to the next step; classifying the second curve group through machine learning algorithms to judge whether there are class problems; if so, extracting the classes with class problems and the corresponding knowledge points; if not, it is determined that there are curriculum problems.
[0047] When the machine learning algorithm for analyzing the first curve group is trained, its input data is the data group corresponding to the learning status sequences of each target object in the class. For example, the data group corresponding to the learning status sequence of the target object can be [A1, A2,..., An], where An represents the learning status label of target object A corresponding to the nth knowledge point; its output data is the label of whether the target object has personal problems. If the label is 0, it means that the corresponding target object does not have personal problems. If the label is 1, it means that the corresponding target object has personal problems.
[0048] When the machine learning algorithm for analyzing the second curve group is trained, its input data is the data group corresponding to the learning status sequences of the target objects in several classes. For example, the data group corresponding to the learning status sequence of the target object in class B can be [AB1, AB2,..., ABn], where ABn represents the learning status label of target object A in class B corresponding to the nth knowledge point; its output data is the label of class problems or the label of curriculum problems. If the label is 1, it means that the corresponding class has problems.
[0049] Exemplarily, for the second set of analysis curves, the output of the algorithm is [C1, C2, …, Cm], where Cm is the output label of the m-th class. If the number of Cm equal to 1 is greater than the set threshold, it is determined that there is no class problem; if the number of Cm equal to 1 is less than or equal to the set threshold, it is determined that the class with Cm equal to 1 has a problem. It should be noted that this set threshold is set according to the number of classes and is default to 80% of the total number of classes.
[0050] Of course, in some other preferred embodiments, the input and output of the machine learning algorithm can also be adjusted to improve efficiency while accurately determining the existing personal problems, class problems or course problems.
[0051] The data used for training the above machine learning algorithm can be extracted and labeled from historical records or obtained from a third-party data platform. The machine learning algorithm is mainly used for classification and can be a neural network model or a support vector machine model.
[0052] If there is a problem with the target object itself, by comparing its learning status sequence with the learning status sequences of other target objects in the same class, the knowledge points that it has not mastered can be determined. If there is a problem with the class, by comparing the learning status sequences of the target objects in this class with the learning status sequences of the target objects in other classes, the knowledge points that the target objects in this class generally have not mastered can be identified; similarly, if there is a problem with the course, the knowledge points that the target objects learning the target course generally have not mastered can also be identified.
[0053] While strengthening the teaching of the knowledge points not mastered, it should also be considered whether it is necessary to strengthen the training of the teaching teacher and whether it is necessary to replace or adjust the target course, so as to make a reliable decision.
[0054] The above embodiments are only used to illustrate the technical method of the present invention and not to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical method of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical method of the present invention.
Claims
1. A multi-dimensional education data analysis and decision support system, characterized in that, It includes an analysis and decision-making module and a connected data processing module; Data processing module: used to collect the behavioral data and learning outcomes of several target objects associated with the target course; among them, several target objects associated with the target course are the applicable objects of the course; and, used to analyze the behavioral data to obtain the concentration labels of the target objects, and parse the learning outcomes to obtain the knowledge mastery sequences of the target objects; among them, the learning outcomes include exam scores and interaction scores; Analysis and decision-making module: used to identify the knowledge distribution sequence in the target course, after time-aligning the knowledge distribution sequence, concentration labels and knowledge mastery sequences, calculate the learning status labels of the target objects according to the concentration labels and the knowledge mastery sequences; and, used to construct the learning status sequence of the target objects according to the learning status labels; jointly judge the learning status sequences of several target objects to determine whether there are preset problems; if so, formulate support measures according to the existing preset problems; among them, the preset problems include individual problems, class problems or course problems.
2. The multi-dimensional education data analysis and decision support system according to claim 1, characterized in that Analyzing the behavioral data to obtain the concentration labels of the target objects includes: extracting the behavioral data of the target objects, and extracting concentration features from the behavioral data; among them, the behavioral data includes video data, and the concentration features include facial features, eye movement features and behavioral features; judging the concentration labels corresponding to the concentration features through machine learning algorithms; among them, the concentration labels correspond to high concentration, medium concentration or low concentration.
3. A multi-dimensional education data analysis and decision support system according to claim 1, characterized in that Parsing the learning outcomes to obtain the knowledge mastery sequences of the target objects includes: identifying several knowledge points examined from the learning outcomes; identifying the mastery situation of the target objects for several knowledge points in the target course from the learning outcomes, setting mastery labels according to the mastery situation, and associating the mastery labels with their corresponding knowledge points.
4. A multi-dimensional education data analysis and decision support system according to claim 1, characterized in that, Identifying the knowledge distribution sequence in the target course includes: extracting the target course, identifying and marking several knowledge points in the target course; matching the associated moments of several knowledge points from the target course, and arranging several knowledge points in the order of the sequence of the associated moments to obtain the knowledge distribution sequence.
5. A multi-dimensional education data analysis and decision support system according to claim 4, characterized in that Time-aligning the knowledge distribution sequence, concentration labels and knowledge mastery sequences includes: extracting several knowledge points in the knowledge distribution sequence and integrating them into a knowledge point group; matching several knowledge points in the knowledge mastery sequence in the knowledge point group, and associating the mastery labels corresponding to the successfully matched knowledge points with their associated moments; matching the concentration labels through the associated moments of several knowledge points in the knowledge point group, and associating the successfully matched concentration labels with the associated moments.
6. The multi-dimensional education data analysis and decision support system according to claim 5, characterized in that, Calculating the learning status labels of the target objects according to the concentration labels and the knowledge mastery sequences includes: extracting the mastery labels and concentration labels associated with the associated moments; matching the learning status labels corresponding to the mastery labels and the concentration labels from a preset label group; among them, the preset label group includes learning status labels corresponding to different combinations of mastery labels and concentration labels.
7. A multi-dimensional education data analysis and decision support system according to claim 1 or 6, characterized in that, Construct the learning status sequence of the target object according to the learning status tags, including: Arrange the correlation moments of several knowledge points in the target course in chronological order to generate a time series; wherein, the correlation moment is the appearance moment of the knowledge point in the target course; Extract the learning status tags of the target object; use the correlation moments in the time series as independent variables and the learning status tags as dependent variables to construct the learning status sequence of the target object.
8. A multi-dimensional education data analysis and decision support system according to claim 7, characterized in that, Jointly judge the learning status sequences of several target objects, including: Integrate the learning status sequences of several target objects into several curve combinations according to a preset question; wherein, the curve combination includes Curve Group 1 and Curve Group 2, Curve Group 1 is used to identify personal problems, and Curve Group 2 is used to identify class problems or course problems; Analyze several curve combinations in the problem recognition order through a machine learning algorithm to determine whether there are preset problems; wherein, the problem recognition order is individual problems, class problems, and course problems in sequence.
9. A multi-dimensional education data analysis and decision support system according to claim 8, characterized in that, Integrate the learning status sequences of several target objects into several curve combinations according to a preset question, including: When the preset question is an individual problem, integrate the learning status sequences of the target objects according to the set class to obtain Curve Group 1; When the preset question is a class problem or a course problem, integrate the learning status sequences of the target objects in different classes to obtain Curve Group 2.
10. A multi-dimensional education data analysis and decision support system according to claim 1, characterized in that, Analyze several curve groups in the problem recognition order through a machine learning algorithm, including: Classify Curve Group 1 through a machine learning algorithm to determine whether there are personal problems; if so, extract the target objects with personal problems and the corresponding knowledge points; if not, proceed to the next step; Classify Curve Group 2 through a machine learning algorithm to determine whether there are class problems; if so, extract the class with class problems and the corresponding knowledge points; if not, determine that there are course problems.