Student personalized education management method based on artificial intelligence

Abstract

The invention belongs to the technical field of big data processing, and provides a student personalized education management method based on artificial intelligence, and the method comprises the steps: analyzing a historical learning record through a recommendation algorithm to generate a recommended course video list of a current student; performing knowledge point target detection on the recommended course video list of the current student to obtain personalized screening courses; updating a recommended course video list of the current student according to the personalized screening course; by intelligently screening big data of all students, the problem of data sparsity of collaborative filtering of a common recommendation algorithm based on implicit feedback can be avoided, the problems of false detection and missing detection of target detection are reduced, and the problem of cold start of the recommendation algorithm is eliminated; and the recommendation accuracy is improved by focusing on the positioning of the video segment.

Description

Technical Field

[0001] The present invention belongs to the technical field of big data processing, and in particular relates to a method for personalized education management of students based on artificial intelligence. Background Art

[0002] The current student personalized education system generally uses artificial intelligence technology to collect, analyze and process students' learning data to identify each student's individual needs, learning style, ability level and interest preferences, thereby tailoring learning content and management strategies for them. This method uses artificial intelligence technology to monitor students' learning progress in real time, dynamically adjust learning paths and resource recommendations, and help teachers and educational institutions provide more accurate support and intervention in the teaching process, promoting students' personalized development and all-round growth.

[0003] The current education is to transform the teacher-led classroom model into a learner-led learning model by using the top of the form. At the same time, online education provides learners with a rich variety of learning resources, such as videos, audios, PPTs and exercises, and learners can choose according to their needs. However, faced with such a large amount of learning resources, learners may feel that the cognitive burden is too heavy.

[0004] Among the existing personalized education methods, the Chinese patent with publication number CN109376269B announces a cross-course video subgraph recommendation method based on graph association. First, the video learning behavior features are extracted through implicit feedback, and a seed video set is first constructed. Then, an association graph between cross-course knowledge points and videos is established, and the association between course videos is calculated using the graph. Finally, combined with the implicit features of user behavior and the seed video set, as well as the association of course videos, cross-course video subgraphs containing knowledge associations are recommended to learners. This method comprehensively considers the user's implicit learning characteristics and the knowledge relevance between course videos, thereby providing learners on the online learning platform with video recommendations that meet their learning preferences and take into account cross-course knowledge connections. However, the personalized education information management method based on learning preferences is the root cause of the "information cocoon". The personalized recommendation algorithm will filter and push information according to the user's preferences, causing the user to be trapped in the content of their interest and difficult to access diverse information, thus forming an "information cocoon". The content accessed is similar to the user's own interests. Due to differences in learners' educational background and abilities, this meets the user's personalized needs, but it may also lead to a narrow field of vision for the user, resulting in unreasonable scheduling of their learning progress, which in turn affects learning outcomes. Summary of the invention

[0005] The purpose of the present invention is to propose a personalized education management method for students based on artificial intelligence to solve one or more technical problems existing in the prior art and at least provide a beneficial choice or create conditions.

[0006] In order to achieve the above object, according to one aspect of the present invention, a method for personalized education management of students based on artificial intelligence is provided, and the method for personalized education management of students based on artificial intelligence comprises the following steps: The server receives the personalized course screening request sent by the client; In response to a personalized course screening request, historical learning records are imported on the server side; the historical learning records are analyzed by a recommendation algorithm to generate a recommended course video list for the current student; Perform knowledge point target detection on the recommended course video list of the current student to obtain personalized screening courses; update the recommended course video list of the current student according to the personalized screening courses; The updated recommended course video list for the current student is sent to the client of the current student.

[0007] The personalized course screening request is request data in JSON format or XML format sent to the server side when the link mark or button of the client application is clicked, and the personalized course screening request at least includes the student's ID identification number; Preferably, the personalized course screening request may also include: subject grades, learning progress, behavioral data, interests and hobbies, study habits, test scores, and classroom performance, wherein the learning progress is the percentage between the number of course videos watched by the student and the total number of course videos.

[0008] Furthermore, the method of importing historical learning records on the server side is: importing historical learning records on the server side, the historical learning records include the cumulative number of times students watch the course video, the cumulative viewing time, and the cumulative number of times students pause and drag the course video.

[0009] Furthermore, the method of analyzing historical learning records through a recommendation algorithm to generate a recommended course video list for the current student is to generate a recommended list of course videos from multiple course videos through a recommendation algorithm based on implicit feedback, and the recommended course video list is a list including one or more course videos.

[0010] Preferably, the recommendation algorithm based on implicit feedback is a seed video set generation method in a cross-course video subgraph recommendation method based on graph association disclosed in the Chinese patent publication with publication number CN109376269B.

[0011] Preferably, the recommendation algorithm based on implicit feedback is the course recommendation method in reference: Zhang Li, Zhang Yiwen, Shan Chunyu, et al. Deep reinforcement learning MOOC course recommendation method based on implicit feedback [J]. Journal of Anqing Normal University (Natural Science Edition), 2024, 30(3): 63-70.

[0012] Among them, the course video is a course teaching video for students to watch.

[0013] Furthermore, the method of performing knowledge point target detection on the recommended course video list of the current student to obtain personalized screening courses includes: Set the similarity threshold and duration threshold. In each course video, if the pixel difference between two adjacent image frames within a period of time is less than the similarity threshold and the duration is greater than the duration threshold, mark the video within this period of time as a knowledge element segment; (Generally, in the course video, there will be no significant pixel difference between two adjacent image frames during the time period of explaining knowledge elements. The pixel difference is the number of different pixels in the two adjacent image frames. The similarity threshold is set to 1 / 10 of the number of pixels in an image frame, and the duration threshold is set to more than 30 seconds); Mark the knowledge meta-segment with the most pauses in each knowledge meta-segment of the course video as the knowledge point segment, and the image frame with the most pauses in the knowledge point segment as the knowledge point target, or mark the region of interest on the image frame with the most pauses in the knowledge point segment as the knowledge point target; The video with the most cumulative pauses or the most cumulative views in the current student's recommended course video list is used as the base video; (when students are learning courses, the image frame position with the most pauses indicates that most people have an interest in the knowledge point or it is difficult to learn the knowledge point. Screening the image frames related to the knowledge point by the number of pauses can extract the video learning behavior feature image frames from the traditional implicit feedback (single or multiple feature ratings) in a more subjective way, and extract the video learning behavior feature image frames more objectively and accurately. Although the passive method slightly increases the computing power consumption of the server (the number of course videos that students need to learn is generally not large, so the computing power consumption cost is not large), this method can improve the accuracy of target detection of related knowledge points in subsequent steps. Screening all courses in a comprehensive target detection manner can eliminate the cold start problem caused by the server's difficulty in accurately calculating the similarity due to the lack of sufficient historical learning records and sufficient rating data for the newly added courses, and reduce the problem of false positives and missed detections (False negatives) of target detection caused by distortion or errors in the feature image frames that target detection relies on).

[0014] The set of course videos except the basic videos in all course videos is taken as the candidate video set; target detection is performed on each course video in the candidate video set, and when the knowledge point target is detected, target tracking and data annotation are performed; the course video in which the knowledge point target appears is taken as the target course video; the video segment in which the knowledge point target appears in the target course video (the video segment is a video clip in which the knowledge point target appears continuously in the target course video) is marked as the knowledge point associated video segment; Let VideoS be the set of video segments associated with knowledge points in the target course video; let VideoST(i) be the number of knowledge meta-segments in the video segment associated with the i-th knowledge point in VideoS, where i is the sequence number; Traverse VideoST(i) in the range of i, and record the sequence number i when VideoST(i) first meets the conditions: VideoST(i)>VideoST(i+1) and VideoST(i)>VideoST(i-1) as the divergence sequence number Divi; record the sequence number i when VideoST(i) first meets the conditions: VideoST(i)<VideoST(i+1) and VideoST(i)<VideoST(i-1) as the convergence sequence number Coni; (The divergence sequence number Divi is the number required when learning the position of this knowledge point in a target course video in the knowledge point-associated video segment. To further learn other related knowledge points, additional learning time is needed; the convergence number Coni is the number of other related knowledge points that do not need to be further learned when learning the position of this knowledge point in a target course video in the knowledge point related video segment, so no additional learning time is needed; the time length between the videos corresponding to these two numbers can indicate the length of time required for students to focus on mastering the corresponding knowledge points in this target course video, and this time length can be used to roughly predict the time cost required for students to master related knowledge points. This time length includes the knowledge elements most relevant to the current knowledge point target); The start time of the video segment associated with the knowledge point corresponding to VideoST(Divi) in VideoS is StartT; the end time of the video segment associated with the knowledge point corresponding to VideoST(Coni) in VideoS is ENDT; the video segment between StartT and ENDT in each target course video is marked as the focused video segment.

[0015] Among them, according to the number of knowledge meta-segments, the focused video segments are located by the divergence number Divi and the convergence number Coni. Intelligent big data screening of all students can avoid the data sparsity problem of collaborative filtering of common recommendation algorithms based on implicit feedback (the data sparsity problem is that students usually only rate a small number of courses, resulting in a very sparse student-course rating matrix), thereby improving the accuracy of the recommendation system.

[0016] Furthermore, the target course videos having the longest focused video segments are used as personalized screening courses.

[0017] Preferably, the method for updating the recommended course video list for the current student according to the personalized screening courses includes: adding the personalized screening courses to the recommended course video list for the current student.

[0018] Preferably, the method for updating the recommended course video list for the current student according to the personalized screening courses includes: deleting the course videos in which the knowledge point objectives do not appear from the recommended course video list for the current student.

[0019] Preferably, the method for updating the recommended course video list of the current student according to the personalized screening course includes: if the value of the current student's learning progress is higher than the average learning progress of all students, the target course video with the longest focused video segment is used as the personalized screening course, and the duration of the focused video segment of the personalized screening course is recorded as GT2, and all the course videos with a total duration of knowledge point segments less than GT2 in the recommended course video list of all current students are deleted, and the personalized screening course is added to the recommended course video list of the current student. (Since the current student's learning progress is fast, in order to enable the current student to quickly learn other related knowledge points when learning this knowledge point position in the target course video in the selected personalized screening course, the student is pushed a course video that can maximize the current student's knowledge breadth in the same learning time).

[0020] Preferably, the method for updating the recommended course video list of the current student according to the personalized screening course includes: if the value of the current student's learning progress is lower than the average learning progress of all students, the target course video with the minimum duration of the focused video segment is the personalized screening course, and the duration of the focused video segment of the personalized screening course is recorded as GT1, and all the course videos with a total duration of knowledge point segments greater than GT1 in the recommended course video list of all current students are deleted, and the personalized screening course is added to the recommended course video list of the current student. (In order to enable the current student to quickly catch up with the learning progress, the target course video with an increasing trend of knowledge points is replaced by the target course video that contains knowledge points and does not need to further learn other related knowledge points when learning to this knowledge point position in the target course video, so that the current student can more easily catch up with the learning progress of other students, and quickly learn according to the course video with the minimum learning time cost to master the missing knowledge points).

[0021] Preferably, an updated list of recommended course videos for the current student is obtained.

[0022] Among them, the above technical scheme, according to the personalized characteristics of students' learning, screens out students with fast learning progress and students with slow learning progress by increasing or decreasing the trend in the video segment containing the knowledge elements most relevant to the current knowledge point goals. If the student's learning progress is slow, the value of the learning progress is low, and if the student's learning progress is fast, the value of the learning progress is high. According to these two situations, students with slow learning progress can learn course videos that focus more on missing knowledge points and quickly catch up with the learning progress, and students with fast learning progress can improve the breadth of knowledge by learning course videos containing more extended knowledge points, break the "information cocoon", and get in touch with diverse knowledge point information to expand the user's knowledge horizon.

[0023] The target detection algorithm is any one of the YOLO target detection algorithm, the R-CNN target detection algorithm, and the Fast R-CNN target detection algorithm.

[0024] Preferably, the method for updating the recommended course video list for the current student according to the personalized screening course includes: recording the number of knowledge meta-segments in the knowledge point associated video segment in the target course video as the knowledge point increment; KL(j) represents the jth knowledge point increment in the target course video, where j is the sequence number of the knowledge point increment; Judge each KL(j) in turn within the range of j: if KL(j) is less than the mean value of the incremental knowledge points in all target course videos and greater than the minimum value of the incremental knowledge points in all target course videos, then add the target course video corresponding to KL(j) to the current student's recommended course video list; if KL(j) is greater than the mean value of the incremental knowledge points in all target course videos, and if the target course video corresponding to KL(j) already exists in the current student's recommended course video list, then delete the target course video corresponding to KL(j) from the current student's recommended course video list to obtain an updated recommended course video list for the current student.

[0025] The above method avoids the problem that the traditional collaborative filtering algorithm based on implicit feedback only tends to recommend popular course videos and rarely recommends unpopular course videos that may be more accurate for students' knowledge points, which improves the accuracy of the recommendation.

[0026] Preferably, the client includes a mobile phone, a tablet computer, a desktop computer or a laptop computer; the client is installed with an APP program, which is used to send a personalized course screening request to the server connected to the client after the student user logs in to the server, and receive a recommended course video list.

[0027] Furthermore, the server side is a UCloud cloud host, an Inspur (INSPUR) NF5280M6 rack-mounted server, or a DELL R740xd rack-mounted server.

[0028] Preferably, the server is a UCloud cloud host. The type of UCloud cloud host is Kuaijie PRO general-purpose cloud host, which uses the following CPUs: AMD EPYC2 (Rome7542) processor and AMD EPYC2 (Rome 7H12) processor; CPU frequency: 2.9GHz-3.4GHz; 2.6GHz-3.3GHz CPU / memory ratio supports 1:2 and 1:4; vCPU number range: 2 cores to 96 cores; The system disk and data disk are cloud storage disks that support RSSD; The intranet bandwidth is 25GbE.

[0029] The present invention also provides a student personalized education information management system based on artificial intelligence, the system comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the computer program running in the following system units: A request sending unit, used for receiving a personalized course screening request sent by a client at the server side; A course generation unit is used to import historical learning records on the server side in response to a personalized course screening request; analyze the historical learning records through a recommendation algorithm to generate a recommended course video list for the current student; The course updating unit is used to perform knowledge point target detection on the recommended course video list of the current student to obtain personalized screening courses; and to update the recommended course video list of the current student according to the personalized screening courses; The course push unit is used to send the updated recommended course video list of the current student to the client of the current student.

[0030] The beneficial effects of the present invention are as follows: the present invention provides a personalized education management method for students based on artificial intelligence, and the beneficial effects include: intelligent screening of big data of all students can avoid the data sparsity problem of collaborative filtering of common recommendation algorithms based on implicit feedback, and reduce the problem of false detection and missed detection of target detection caused by distortion or errors in feature image frames relied on by target detection; screening all courses in a comprehensive target detection manner can eliminate the cold start problem caused by the server having difficulty in accurately calculating the similarity of newly added courses due to insufficient historical learning records and sufficient rating data; focusing on the positioning of video segments avoids the traditional collaborative filtering algorithm based on implicit feedback that only tends to recommend popular course videos, and recommends less unpopular course videos that may be more accurate for student knowledge points, thereby improving the accuracy of recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The above and other features of the present invention will become more obvious through a detailed description of the embodiments shown in the accompanying drawings. The same reference numerals in the accompanying drawings of the present invention represent the same or similar approximate completion items. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those of ordinary skill in the art, other accompanying drawings can be obtained based on these accompanying drawings without creative work. In the accompanying drawings: Figure 1 Shown is a flow chart of a method for personalized education management of students based on artificial intelligence; Figure 2 Shown is the structural diagram of the student personalized education information management system based on artificial intelligence. DETAILED DESCRIPTION

[0032] The following will be combined with the embodiments and drawings to clearly and completely describe the concept, specific structure and technical effects of the present invention, so as to fully understand the purpose, scheme and effect of the present invention. It should be noted that the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0033] Example 1 like Figure 1 The flowchart of the method for personalized education management of students based on artificial intelligence is shown below. Figure 1 To illustrate the student personalized education management method based on artificial intelligence according to an embodiment of the present invention, the method comprises the following steps: The server receives the personalized course screening request sent by the client; In response to a personalized course screening request, historical learning records are imported on the server side; the historical learning records are analyzed by a recommendation algorithm to generate a recommended course video list for the current student; Perform knowledge point target detection on the recommended course video list of the current student to obtain personalized screening courses; update the recommended course video list of the current student according to the personalized screening courses; The updated recommended course video list for the current student is sent to the client of the current student.

[0034] The personalized course screening request is request data in JSON format sent to the server when a button of the client application is clicked, and the personalized course screening request at least includes the student's ID identification number; Preferably, the personalized course screening request may also include: subject grades, learning progress, behavioral data, interests and hobbies, study habits, test scores, and classroom performance, wherein the learning progress is the percentage between the number of course videos watched by the student and the total number of course videos.

[0035] Furthermore, the method of importing historical learning records on the server side is: importing historical learning records on the server side, the historical learning records include the cumulative number of times students watch the course video, the cumulative viewing time, and the cumulative number of times students pause and drag the course video.

[0036] Furthermore, the method of analyzing historical learning records through a recommendation algorithm to generate a recommended course video list for the current student is to generate a recommended list of course videos from multiple course videos through a recommendation algorithm based on implicit feedback, and the recommended course video list is a list including one or more course videos.

[0037] Preferably, the recommendation algorithm based on implicit feedback is a seed video set generation method in a cross-course video subgraph recommendation method based on graph association disclosed in the Chinese patent publication with publication number CN109376269B.

[0038] Among them, the course video is a course teaching video for students to watch.

[0039] Furthermore, the method of performing knowledge point target detection on the recommended course video list of the current student to obtain personalized screening courses includes: Set a similarity threshold and a duration threshold. In each course video, if the pixel difference between two adjacent image frames within a period of time is less than the similarity threshold and the duration is greater than the duration threshold, mark the video within this period of time as a knowledge segment. Among them, the similarity threshold is set to 1 / 10 of the number of pixels in an image frame, and the duration threshold is set to more than 30 seconds; Annotate the region of interest on the image frame with the most pauses in the knowledge point segment as the knowledge point target; The basic video is the course video with the highest cumulative pause times or cumulative viewing times in the recommended course video list of the current student; The set of course videos except the basic videos in all course videos is taken as the candidate video set; target detection is performed on each course video in the candidate video set, and when the knowledge point target is detected, target tracking and data annotation are performed; the course video in which the knowledge point target appears is taken as the target course video; the video segment in which the knowledge point target appears in the target course video is marked as the knowledge point associated video segment; Let VideoS be the set of video segments associated with knowledge points in the target course video; let VideoST(i) be the number of knowledge meta-segments in the video segment associated with the i-th knowledge point in VideoS, where i is the sequence number; Traverse VideoST(i) in the range of i in sequence, and record the sequence number i when VideoST(i) first meets the conditions: VideoST(i)>VideoST(i+1) and VideoST(i)>VideoST(i-1) as the divergence sequence number Divi; record the sequence number i when VideoST(i) first meets the conditions: VideoST(i)<VideoST(i+1) and VideoST(i)<VideoST(i-1) as the convergence sequence number Coni; The start time of the video segment associated with the knowledge point corresponding to VideoST(Divi) in VideoS is StartT; the end time of the video segment associated with the knowledge point corresponding to VideoST(Coni) in VideoS is ENDT; the video segment between StartT and ENDT in each target course video is marked as the focused video segment.

[0040] Among them, the key source code of the Java language description of performing knowledge point target detection on the recommended course video list of the current student to obtain personalized screening courses is: class VideoFrame { private int[][]pixels; / / Pixels of the image frame private long timestamp; / / Timestamp of image frame public class VideoAnalysis { private static final int TS_FACTOR = 10; / / Similarity threshold factor private static final int TT_THRESHOLD = 30; / / Duration threshold (seconds) public List <knowledgesegment>identifyKnowledgeSegments(List <videoframe>frames) { List <knowledgesegment>knowledgeSegments = new ArrayList<>(); int similarityThreshold = frames.get(0).getPixels().length / TS_FACTOR; for (int i = 1; i < frames.size(); i++) { VideoFrame currentFrame = frames.get(i); VideoFrame previousFrame = frames.get(i - 1); int pixelDifference = calculatePixelDifference(previousFrame.getPixels(), currentFrame.getPixels()); if (pixelDifference < similarityThreshold) { / / Check the duration if (i > 0 && (currentFrame.getTimestamp() - previousFrame.getTimestamp()) > TT_THRESHOLD) { knowledgeSegments.add(new KnowledgeSegment(previousFrame.getTimestamp(), currentFrame.getTimestamp())); }}} return knowledgeSegments;} private int PixelDifference(int[][] pixels1, int[][] pixels2) { / / Implement pixel difference calculation int difference = 0; for (int i = 0; i < pixels1.length; i++) { for (int j = 0; j < pixels1[i].length; j++) { difference += Math.abs(pixels1[i][j]- pixels2[i][j]); }}return difference; } class KnowledgeSegment { private long startTime; private long endTime; public KnowledgeSegment(long startTime, long endTime) { this.startTime = startTime; this.endTime = endTime;} public long getStartTime() { return startTime;} public long getEndTime() { return endTime;}} public class VideoAnalysis { / / Collection of video segments associated with knowledge points in the target course video private List <knowledgesegment>videoS; public VideoAnalysis(List <knowledgesegment>videoS) { this.videoS = videoS;} public void analyzeKnowledgeSegments() { int[] videoSTCounts = new int[videoS.size()]; / / Store the number of knowledge segments in the video segment associated with each knowledge point integer divi = null; / / divergent sequence number integer coni = null; / / convergence number for (int i = 1; i <videoSTCounts.length - 1; i++) { if (videoSTCounts[i]>videoSTCounts[i + 1]&&videoSTCounts[i]>videoSTCounts[i - 1]) { if (divi == null) { / / Find the first divergent sequence number divi = i; }} else if (videoSTCounts[i] <videoSTCounts[i + 1]&&videoSTCounts[i]<videoSTCounts[i - 1]) { if (coni == null) { / / Find the first convergence number coni = i; }}} if (divi != null&&coni != null) { long startT = videoS.get(divi).getStartTime(); long endT = videoS.get(coni).getEndTime(); markFocusedSegment(startT, endT); } else { System.out.println("No divergence or convergence sequence number found."); } }.

[0041] Furthermore, the target course videos having the longest focused video segments are used as personalized screening courses.

[0042] Preferably, the method for updating the recommended course video list for the current student according to the personalized screening courses includes: adding the personalized screening courses to the recommended course video list for the current student.

[0043] Preferably, an updated list of recommended course videos for the current student is obtained.

[0044] Among them, the target detection algorithm is the YOLO target detection algorithm.

[0045] Preferably, the client includes a mobile phone, a tablet computer, a desktop computer or a laptop computer; the client is installed with an APP program, which is used to send a personalized course screening request to the server connected to the client after the student user logs in to the server, and receive a recommended course video list.

[0046] Preferably, the server is a UCloud cloud host. The type of UCloud cloud host is Kuaijie PRO general-purpose cloud host, which uses the following CPUs: AMD EPYC2 (Rome7542) processor and AMD EPYC2 (Rome 7H12) processor; CPU frequency: 2.9GHz-3.4GHz; 2.6GHz-3.3GHz CPU / memory ratio supports 1:2 and 1:4; vCPU number range: 2 cores to 96 cores; The system disk and data disk are cloud storage disks that support RSSD; The intranet bandwidth is 25GbE.

[0047] Example 2 This embodiment 2 replaces the following contents on the basis of embodiment 1: The region of interest is marked on the image frame with the most pauses in the knowledge point segment and replaced as the knowledge point target. Specifically: the knowledge meta-segment with the most pauses in each knowledge meta-segment of the course video is marked as the knowledge point segment, and the image frame with the most pauses in the knowledge point segment is taken as the knowledge point target.

[0048] Preferably, the method for updating the recommended course video list for the current student according to the personalized screening courses includes: deleting the course videos in which the knowledge point objectives do not appear from the recommended course video list for the current student.

[0049] Preferably, the method for updating the recommended course video list for the current student based on the personalized screening courses is replaced by: if the value of the current student's learning progress is higher than the average learning progress of all students, the target course video with the maximum duration of the focused video segment is used as the personalized screening course, and the duration of the focused video segment of the personalized screening course is recorded as GT2, and all course videos with a total duration of all knowledge point segments less than GT2 in the recommended course video list of all current students are deleted, and the personalized screening courses are added to the recommended course video list of the current student.

[0050] Among them, the key source code of the Java language description of the method for updating the recommended course video list of the current student according to the personalized screening course in Example 2 is: public long getDuration() { return endTime - startTime; / / Get the duration of the focused video segment } class CourseVideo { / / Get recommended course videos private String title; private List <focusedsegment>focusedSegments; public CourseVideo(String title) { this.title = title; this.focusedSegments = new ArrayList<>();} public void addFocusedSegment(FocusedSegment segment) { focusedSegments.add(segment);} public long getTotalKnowledgeDuration() { long totalDuration = 0; for (FocusedSegment segment : focusedSegments) { totalDuration += segment.getDuration();} return totalDuration; / / 获取知识点段的总时长 } class Student { private String name; private double learningProgress; / / 学生的学习进度 private List <coursevideo>recommendedVideos; public Student(String name, double learningProgress) { this.name = name; this.learningProgress = learningProgress; this.recommendedVideos = new ArrayList<>(); } public double getLearningProgress() { return learningProgress;} public List <coursevideo>getRecommendedVideos() { return recommendedVideos;} public void addRecommendedVideo(CourseVideo video) { recommendedVideos.add(video);} public class CourseRecommendation2 { public static void updateRecommendedCourses(Student student, List <student>allStudents) { double averageProgress = calculateAverageLearningProgress(allStudents); if (student.getLearningProgress()>averageProgress) { CourseVideo personalizedCourse = findMaxDurationFocusedCourse(student.getRecommendedVideos()); if (personalizedCourse != null) { long GT2 = personalizedCourse.getTotalKnowledgeDuration(); / / Get the total duration of the focused video segment of the personalized screening course / / Delete all the course videos in the recommended course list whose total duration of knowledge points is less than GT2 student.getRecommendedVideos().removeIf(video ->video.getTotalKnowledgeDuration() <GT2); / / Add personalized filtered courses to the recommendation list and update recommended courses student.addRecommendedVideo(personalizedCourse);}}} private static double calculateAverageLearningProgress(List <student>students) { double totalProgress = 0.0; for (Student student : students) { totalProgress += student.getLearningProgress(); }return totalProgress / students.size();} private static CourseVideo findMaxDurationFocusedCourse(List <coursevideo>videos) CourseVideo maxCourse = null; for (CourseVideo video : videos) { if (maxCourse == null || video.getTotalKnowledgeDuration()>maxCourse.getTotalKnowledgeDuration()) { maxCourse = video; / / Find the target course video with the longest focused video segment }}return maxCourse; }.

[0051] Example 3 In this embodiment 3, the following contents are replaced on the basis of embodiment 1: Preferably, the method for updating the recommended course video list for the current student based on the personalized screening courses is replaced by: if the value of the current student's learning progress is lower than the average learning progress of all students, the target course video with the minimum duration of the focused video segment is used as the personalized screening course, and the duration of the focused video segment of the personalized screening course is recorded as GT1, and all course videos with a total duration of all knowledge point segments greater than GT1 in the recommended course video list of all current students are deleted, and the personalized screening courses are added to the recommended course video list of the current student.

[0052] Among them, the key source code of the Java language description of the method for updating the recommended course video list of the current student according to the personalized screening course in Example 3 is: public class CourseRecommendation3 { public static void updateRecommendedCourses(Student student, List <student>allStudents) { double averageProgress = calculateAverageLearningProgress(allStudents); if (student.getLearningProgress() < averageProgress) { / / Learning progress is below the average CourseVideo personalizedCourse = findMinDurationFocusedCourse(student.getRecommendedVideos()); if (personalizedCourse != null) { long GT1 = personalizedCourse.getMinFocusedSegmentDuration(); / / Get the minimum focused video segment duration of the personalized filtered course / / Delete all course videos in the recommended course list whose total knowledge segment duration is greater than GT1 student.getRecommendedVideos().removeIf(video -> video.getTotalKnowledgeDuration() > GT1); / / Add the personalized filtered course to the recommended list student.addRecommendedVideo(personalizedCourse);}}} private static CourseVideo findMinDurationFocusedCourse(List <coursevideo>videos) CourseVideo minCourse = null; for (CourseVideo video : videos) { if (minCourse == null || video.getMinFocusedSegmentDuration() <minCourse.getMinFocusedSegmentDuration()) { minCourse = video; / / Find the target course video with the minimum focused video segment duration }}return minCourse; / / Return the found course video }.

[0053] Example 4 In this embodiment 4, the following contents are replaced on the basis of embodiment 1: Preferably, the method for updating the recommended course video list for the current student according to the personalized screening course is replaced by: recording the number of knowledge meta-segments in the knowledge point associated video segment in the target course video as the knowledge point increment; KL(j) represents the jth knowledge point increment in the target course video, where j is the sequence number of the knowledge point increment; Judge each KL(j) in turn within the range of j: if KL(j) is less than the mean value of the incremental knowledge points in all target course videos and greater than the minimum value of the incremental knowledge points in all target course videos, then add the target course video corresponding to KL(j) to the current student's recommended course video list; if KL(j) is greater than the mean value of the incremental knowledge points in all target course videos, and if the target course video corresponding to KL(j) already exists in the current student's recommended course video list, then delete the target course video corresponding to KL(j) from the current student's recommended course video list to obtain an updated recommended course video list for the current student.

[0054] Among them, the key source code of the Java language description of the method for updating the recommended course video list of the current student according to the personalized screening course in Example 4 is: class CourseVideo { private String title; private int knowledgePointIncrement; / / Knowledge point increment public CourseVideo(String title, int knowledgePointIncrement) { this.title = title; this.knowledgePointIncrement = knowledgePointIncrement;}} public int getKnowledgePointIncrement() { return knowledgePointIncrement;}} public List <coursevideo>getRecommendedVideos() { return recommendedVideos;} public void addRecommendedVideo(CourseVideo video) { recommendedVideos.add(video);} public void removeRecommendedVideo(CourseVideo video) { recommendedVideos.remove(video);}} public class VideoRecommendation4 { public static void updateRecommendedCourses(Student student, List <coursevideo>targetVideos) { double averageIncrement = calculateAverageKnowledgePointIncrement(targetVideos); int minIncrement = findMinKnowledgePointIncrement(targetVideos); for (CourseVideo video : targetVideos) { int increment = video.getKnowledgePointIncrement(); / / Determine KL(j) if (increment<averageIncrement&&increment> minIncrement) { / / If KL(j) is between the mean and the minimum, add it to the recommendation list student.addRecommendedVideo(video); } else if (increment>averageIncrement) { / / If KL(j) is greater than the mean and the video is already in the recommendation list, delete it if (student.getRecommendedVideos().contains(video)) { student.removeRecommendedVideo(video); }}}} private static double calculateAverageKnowledgePointIncrement(List <coursevideo>videos) { double total = 0.0; for (CourseVideo video : videos) { total += video.getKnowledgePointIncrement();} return total / videos.size();} private static int findMinKnowledgePointIncrement(List <coursevideo>videos) int min = Integer.MAX_VALUE; for (CourseVideo video : videos) { if (video.getKnowledgePointIncrement() <min) { min = video.getKnowledgePointIncrement(); } }return min; / / Return the minimum knowledge point increment }.

[0055] The embodiment of the present invention provides a student personalized education information management system based on artificial intelligence, such as Figure 2 Shown is a structural diagram of the student personalized education information management system based on artificial intelligence of the present invention. The student personalized education information management system based on artificial intelligence of this embodiment includes: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the above-mentioned embodiment of the student personalized education information management system based on artificial intelligence are implemented.

[0056] The system comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to run in the following units of the system: A request sending unit, used for receiving a personalized course screening request sent by a client at the server side; A course generation unit is used to import historical learning records on the server side in response to a personalized course screening request; analyze the historical learning records through a recommendation algorithm to generate a recommended course video list for the current student; The course updating unit is used to perform knowledge point target detection on the recommended course video list of the current student to obtain personalized screening courses; and to update the recommended course video list of the current student according to the personalized screening courses; The course push unit is used to send the updated recommended course video list of the current student to the client of the current student.

[0057] The student personalized education information management system based on artificial intelligence can be run on computing devices such as desktop computers, notebooks, PDAs and cloud servers. The student personalized education information management system based on artificial intelligence can be operated on systems that may include, but are not limited to, processors and memories. Those skilled in the art will appreciate that the examples are merely examples of student personalized education information management systems based on artificial intelligence, and do not constitute limitations on student personalized education information management systems based on artificial intelligence. They may include more or fewer components than the examples, or combine certain components, or different components. For example, the student personalized education information management system based on artificial intelligence may also include input and output devices, network access devices, buses, and the like.

[0058] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the operation system of the student personalized education information management system based on artificial intelligence, and uses various interfaces and lines to connect the various parts of the entire operation system of the student personalized education information management system based on artificial intelligence.

[0059] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the student personalized education information management system based on artificial intelligence by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (SmArray1t Media Carray1d, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Carray1d), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0060] Although the description of the present invention has been quite detailed and has been described in particular with respect to several described embodiments, it is not intended to be limited to any of these details or embodiments or any particular embodiment, so as to effectively cover the intended scope of the present invention. In addition, the present invention is described above with the embodiments foreseeable by the inventors, and its purpose is to provide a useful description, and those non-substantial changes to the present invention that are not currently foreseen may still represent equivalent changes of the present invention.< / coursevideo> < / coursevideo> < / coursevideo> < / coursevideo> < / coursevideo> < / student> < / coursevideo> < / student> < / student> < / coursevideo> < / coursevideo> < / focusedsegment> < / knowledgesegment> < / knowledgesegment> < / knowledgesegment> < / videoframe> < / knowledgesegment>

Claims

1. A personalized student education management method based on artificial intelligence, characterized in that: The method comprises the following steps: The server receives the personalized course screening request sent by the client; In response to a personalized course screening request, historical learning records are imported on the server side; the historical learning records are analyzed by a recommendation algorithm to generate a recommended course video list for the current student; Perform knowledge point target detection on the recommended course video list of the current student to obtain personalized screening courses; update the recommended course video list of the current student according to the personalized screening courses; The updated recommended course video list for the current student is sent to the client of the current student.

2. The method for personalized education management of students based on artificial intelligence according to claim 1 is characterized in that: The method for importing historical learning records on the server side is: importing historical learning records on the server side, the historical learning records include the cumulative number of times students have watched the course video, the cumulative viewing time, and the cumulative number of times students have paused and dragged the course video.

3. The method for personalized education management of students based on artificial intelligence according to claim 1 is characterized in that: The method for performing knowledge point target detection on the recommended course video list of the current student to obtain personalized screening courses includes: setting a similarity threshold and a duration threshold, and in each course video, if the pixel difference between two adjacent image frames within a period of time is less than the similarity threshold and the time length is greater than the duration threshold, then marking the video within this period of time as a knowledge meta-segment; Mark the knowledge meta-segment with the most pauses in each knowledge meta-segment of the course video as the knowledge point segment, and the image frame with the most pauses in the knowledge point segment as the knowledge point target, or mark the region of interest on the image frame with the most pauses in the knowledge point segment as the knowledge point target; The basic video is the course video with the highest cumulative pause times or cumulative viewing times in the recommended course video list of the current student; The set of course videos excluding basic videos in all course videos is taken as the candidate video set; target detection is performed on each course video in the candidate video set, and when a knowledge point target is detected, target tracking and data labeling are performed; the course video in which the knowledge point target appears is taken as the target course video; the video segment in which the knowledge point target appears in the target course video is marked as the knowledge point associated video segment.

4. The method for personalized education management of students based on artificial intelligence according to claim 3 is characterized in that: Let VideoS be the set of video segments associated with knowledge points in the target course video; let VideoST(i) be the number of knowledge meta-segments in the video segment associated with the i-th knowledge point in VideoS, where i is the sequence number; Traverse VideoST(i) in the range of i in sequence, and record the sequence number i when VideoST(i) first meets the conditions: VideoST(i)>VideoST(i+1) and VideoST(i)>VideoST(i-1) as the divergence sequence number Divi; record the sequence number i when VideoST(i) first meets the conditions: VideoST(i)<VideoST(i+1) and VideoST(i)<VideoST(i-1) as the convergence sequence number Coni; The start time of the video segment associated with the knowledge point corresponding to VideoST(Divi) in VideoS is defined as StartT; the end time of the video segment associated with the knowledge point corresponding to VideoST(Coni) in VideoS is defined as ENDT; the video segment between StartT and ENDT in each target course video is marked as a focused video segment; The target course videos with the longest focused video segments are used as personalized filtering courses.

5. The method for personalized education management of students based on artificial intelligence according to claim 4 is characterized in that: The method for updating the recommended course video list for the current student according to the personalized screening courses includes: adding the personalized screening courses to the recommended course video list for the current student.

6. The method for personalized education management of students based on artificial intelligence according to claim 4 is characterized in that: The method for updating the recommended course video list for the current student according to the personalized screening courses includes: deleting the course videos in which the knowledge point objectives do not appear from the recommended course video list for the current student.

7. The method for personalized education management of students based on artificial intelligence according to claim 4 is characterized in that: The method for updating the recommended course video list for the current student based on the personalized screening courses is replaced by: if the current student's learning progress value is higher than the average learning progress of all students, the duration of the focused video segment of the personalized screening course is recorded as GT2, and all course videos with a total duration of knowledge point segments less than GT2 in the recommended course video list of all current students are deleted, and the personalized screening courses are added to the recommended course video list of the current student.

8. The method for personalized education management of students based on artificial intelligence according to claim 4 is characterized in that: The method for updating the recommended course video list for the current student based on personalized screening courses is replaced by: if the current student's learning progress value is lower than the average learning progress of all students, the personalized screening course is replaced by: the target course video with the minimum duration of the focused video segment as the personalized screening course; the duration of the focused video segment of the personalized screening course is recorded as GT1, and all course videos with a total duration of knowledge point segments greater than GT1 in the recommended course video list of all current students are deleted, and the personalized screening course is added to the recommended course video list of the current student.

9. The method for personalized education management of students based on artificial intelligence according to claim 4 is characterized in that: The method for updating the recommended course video list for the current student based on the personalized screening course is replaced by: the number of knowledge meta-segments in the knowledge point associated video segment in the target course video is recorded as the knowledge point increment; KL(j) represents the jth knowledge point increment in the target course video, where j is the sequence number of the knowledge point increment; Judge each KL(j) in the range of j: if KL(j) is less than the mean value of the knowledge point increments in all target course videos and greater than the minimum value of the knowledge point increments in all target course videos, then add the target course video corresponding to KL(j) to the recommended course video list of the current student; If KL(j) is greater than the mean of the incremental knowledge points in all target course videos, and if the target course video corresponding to KL(j) already exists in the recommended course video list of the current student, then the target course video corresponding to KL(j) is deleted from the recommended course video list of the current student to obtain an updated recommended course video list for the current student.

10. The student personalized education information management system based on artificial intelligence is characterized by: The student personalized education information management system based on artificial intelligence includes: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the student personalized education management method based on artificial intelligence described in any one of claims 1-9 are implemented.

Images