Intelligent teaching data acquisition and processing intelligent control system

Abstract

The application discloses a kind of intelligent teaching data acquisition processing intelligent control system, it is related to electric teaching device technical field, and is used for the teaching organization control of large class interactive classroom.System includes the student station, teacher end control device, data processing device and grouping instruction device connected by communication interface;Student station collects response event and role event, and the mutual teaching correlation between the explanation student and the student needing help is registered in role event;Data processing device takes classroom link as boundary, assembles previous measurement response, mutual teaching correlation and post-test response into event pair and is screened by quality gate, and the score change of student needing help before and after measurement is attributed to mutual teaching gain of explanation student, and effective explainer is marked accordingly, and grouping instruction is generated according to the constraint of at least one effective explainer per group in link gap, and group is changed by grouping instruction device and indication unit execution.Making mutual teaching effectiveness become re-calculation control quantity, it is beneficial to person by person leaving trace and mutual teaching resource balanced allocation.

Description

Technical Field

[0001] This invention relates to the field of electric teaching devices, specifically an intelligent control system for intelligent teaching data acquisition and processing. Background Technology

[0002] In large, interactive classroom settings for teacher training courses in universities, teaching organization has shifted from one-way lectures to a multi-stage model primarily based on student self-directed learning. A typical class includes pre-class assessment, explanation, group tasks, peer teaching, and peer evaluation. Teachers need to track the participation and performance of dozens of students in each stage and organize learning groups accordingly. Currently, teachers score student performance based on memory after class and record it in paper forms. Learning groups are arranged according to a list determined at the beginning of the semester, with minimal adjustments throughout the semester; occasional reshuffling relies on manual work by the teacher. The cost of this approach is that individual, stage-by-stage interaction is difficult to document in class; post-class recall records are incomplete and delayed; group lists are not adjusted based on classroom performance throughout the semester, making it impossible to grasp the distribution of explanation abilities within groups; and there is a lack of criteria for judging which student's explanation facilitated the understanding of their peers during peer teaching.

[0003] To improve the recording of classroom interaction, one type of smart teaching tool uses students' mobile terminals to push classroom exercises, provide timed responses, statistical analysis of answers, and random roll call, offering three grouping methods: system-random, student-selective, and teacher-assigned. However, the statistical results of these tools are only visible to teachers; there is no linkage between the grouping method and the statistical data, and the grouping does not reflect students' actual performance in class. Another representative existing technology further utilizes the collected data for automatic grouping. For example, Chinese patent application CN110867103A discloses an intelligent dynamic grouping learning system based on personalized student learning. This system uses a cloud server to collect data on students' time spent answering questions, accuracy, performance in collaborating with classmates, and enthusiasm for answering questions, establishing ability evaluation indices, collaboration preference indices, and activity indices. After the teacher initiates a grouping request, the system automatically divides students into learning groups according to question type and grouping parameters. The shortcomings of this type of scheme are as follows: All its indices are based on the frequency and accuracy of individual student behaviors, reflecting how many answers a student gave and how well they answered, but failing to measure the effectiveness of peer teaching. Students who frequently explain but whose peers do not improve are indistinguishable from quiet students who provide effective explanations. Grouping results do not include constraints on the allocation of explanation abilities within the group, failing to guarantee that each group has an effective supply of explanations. Grouping is triggered by teacher requests and is not adjusted dynamically with each class session. Another type of dynamic group teaching system maps students' learning outcomes to the next stage of teaching content according to preset threshold and proportion rules, achieving the distribution of learning content, but similarly does not involve measuring peer teaching relationships or controlling group composition.

[0004] Therefore, how to collect interactive data from each student during classroom sessions, transform the effectiveness of peer teaching into a recalcible control quantity, and automatically maintain the supply of instruction for each learning group between classroom sessions has become a technical problem that needs to be solved in this field. There is an urgent need for a smart teaching data collection and processing intelligent control system for large-class interactive classrooms. Summary of the Invention

[0005] To address the technical problems in existing large-class interactive classrooms, such as the lack of individual interaction record keeping, the inability to measure the effectiveness of peer teaching, and the lack of adjustment of group composition according to classroom performance, the present invention aims to provide an intelligent teaching data acquisition and processing control system. This system collects interactive data from each student during classroom sessions, quantifies the effectiveness of peer teaching, and automatically generates and executes group adjustments between classroom sessions based on this data, ensuring that each learning group continuously receives data-verified instruction.

[0006] The overall technical concept of this invention is to use "which student is teaching, who is being taught, and whether the teaching is effective" as the main data for classroom control. Student stations are equipped with role keys; students' button presses to request explanations or help directly register the mutual teaching relationship between the explaining student and the assisted student within the system, transforming mutual teaching from unobservable verbal activities into time-stamped hardware events. The teacher-side control device sends segment boundary signals according to the classroom script, providing segment-level time coordinates for all events. Based on this, the data processing device assembles two responses from the assisted student on the same knowledge point before and after the establishment of the mutual teaching relationship into event pairs. The first response acts as a pre-test, and the second response as a post-test; the change in scores between the two responses is the direct measurable output of the mutual teaching effect. The event pairs are first subjected to a quality gate to eliminate samples with timeout windows and no room for improvement, and small sample gains are pushed into low confidence levels. The change is then attributed to the corresponding explaining student, accumulating to the mutual teaching gain for that explaining student. Students whose peer-learning gains reach the gain threshold are marked as effective presenters. Group control then uses "at least one effective presenter per learning group" as a hard condition for generation. During class breaks, the process is recalculated by the boundary signal of the learning session. Grouping instructions are executed on the spot via the grouping indicator device and the student station indicator unit. Data collection, attribution, judgment, grouping, and execution cycle with each class session. The allocation of peer-learning resources is adjusted step by step according to the actual classroom performance. All intermediate data is stored in the student's growth portfolio step by step.

[0007] To achieve the above objectives, the present invention provides an intelligent control system for intelligent teaching data acquisition and processing, comprising at least two student stations, a teacher-end control device, a data processing device, and a grouping indicator device. The student stations, the teacher-end control device, and the grouping indicator device are respectively connected to the data processing device via communication interfaces.

[0008] The student station is equipped with response key groups, role keys, and indicator units. The response key groups collect student response events during class sessions, including student station identifier, session identifier, question identifier, response result, and response delay. The role keys collect role events, including requests for explanation and requests for help. Requests for explanation events register the mutual teaching relationship between the explaining student and the student receiving help. The teacher-side control device sends session boundary signals to the data processing device, marking the type and start / end times of each class session.

[0009] The data processing device receives response events, role events, and segment boundary signals. Within the classroom segment sequence of the same lesson, it assembles the response events of students who have received assistance before the establishment of the mutual teaching relationship for questions of the same knowledge point into pre-test responses and response events that have been received after the establishment of the mutual teaching relationship into post-test responses. These responses are then combined with corresponding role events to form event pairs. A quality gate screening is performed on these event pairs, which includes at least a lower limit for the number of student samples and an upper limit for the time window from the establishment of the mutual teaching relationship to the post-test response. For event pairs that pass the quality gate screening, the data processing device attributes the change in the student's post-test response score relative to the pre-test response score to the corresponding instructor, calculates the instructor's mutual teaching gain, and marks the instructor as a valid instructor when the instructor's mutual teaching gain reaches a gain threshold.

[0010] During class breaks, the data processing device is triggered by the boundary signal of the session and generates grouping instructions according to the grouping constraints. The grouping constraints include at least one effective lecturer for each learning group. The grouping instructions are output to the grouping indicator device and the indicator unit of the student station, which execute the instructions to guide the students to adjust their learning groups according to the grouping instructions.

[0011] Furthermore, the data processing device performs weighted statistics on the number of interactive events for each student in the already conducted class sessions according to the preset scale weights, and obtains the student's participation level. Interactive events include at least response events, requests for explanations, and submission of peer reviews.

[0012] Furthermore, the gain threshold is taken as the smaller value between the upper quartile of the peer teaching gain distribution of the class in the current lesson and the preset baseline value. When the smaller value is less than zero, the gain threshold is zero, and it is updated on a rolling basis with the classroom activities, so that the threshold can adapt to both the overall level of the class and the value-added baseline.

[0013] Furthermore, the grouping constraints also include that the range of the average participation rate among learning groups is no greater than a preset equilibrium threshold, and that the mutual teaching pairing of the same instructor and the same assisted student is not repeated in two adjacent rounds of grouping; when generating grouping instructions, effective instructors are rotated and assigned to each learning group in descending order of mutual teaching gain, and then the remaining students are added to each learning group in a serpentine order according to participation.

[0014] Furthermore, the data processing device binds confidence markers to the mutual teaching gain. The confidence marker is high when the number of students receiving assistance is not lower than the lower limit of the sample size, and low in other cases. Mutual teaching gains with low confidence markers do not participate in the determination of effective instructor labeling and grouping constraints, thereby isolating the gain and grouping decision under small sample conditions.

[0015] Furthermore, when a student's participation rate is lower than the lower quartile of the class participation rate distribution for that class, the data processing device sends a downgraded task package to the student's student station; when a student's participation rate is not lower than the upper quartile of the class participation rate distribution for that class and the student has not been marked as a valid instructor, the data processing device sends an application prompt for an instructor to the student's student station.

[0016] Furthermore, when any one of the following three events occurs: the peer teaching gain of the instructor reaches the gain threshold for the first time, the student's participation level exceeds the upper quartile of the class participation distribution for the first time, or the assisted student's post-test response score is higher than the pre-test response score for the first time, the data processing device sends an incentive flag change instruction to the instruction unit of the corresponding student station and pushes a prompt message to the teacher's control device.

[0017] Furthermore, when communication between the student station and the data processing device is interrupted, the student station caches the response event and role event in the local storage module. After communication is restored, the response event is sent back with a timestamp. The data processing device recalculates the mutual teaching gain according to the step boundaries. The recalculation result is only used to update the growth profile of the student's data for each step and the input of the next round of grouping. When the number of effective instructors is less than the number of learning groups, the data processing device uses the instructor configuration of the missing learning groups from the previous round of grouping and sends a manual assignment prompt to the teacher control device.

[0018] Furthermore, the data processing device writes each student's participation, peer learning gains, event pair details, and confidence markers for each class session into the growth portfolio storage module. The growth portfolio is accumulated by class and can be accessed by the teacher's control device and the supervisor's evaluation terminal.

[0019] Compared with the prior art, the beneficial effects of the present invention are as follows.

[0020] Existing classroom systems that statistically analyze individual behavior frequency and accuracy can only answer how many questions a student answered and how well they answered, leaving a gap in data on the crucial interactive classroom element of peer teaching. This invention uses role keys to register requests for explanations and assistance as hardware events. Then, it pairs pre-test responses, peer teaching relationships, and post-test responses into event pairs, using classroom stages as boundaries. The change in the assisted student's score between the pre-test and post-test is directly included in the peer teaching gain under the explaining student's name. This effectively determines the effectiveness of peer teaching, generating data within the system that can be collected, compared, and directly used for group control—distinguishing between students who frequently explain but whose peers do not improve, and students who speak less but whose explanations are effective, based on the gain data.

[0021] The small sample size and high noise levels in classrooms make differential statistics highly susceptible to random fluctuations. A quality gate eliminates events where the interval between the establishment of the peer-teaching association and the post-test exceeds the upper limit of the time window, ensuring the post-test occurs immediately after the explanation. Gains with fewer assisted samples than the lower limit are still calculated but marked as low confidence and not included in the valid explainer label. Samples that have nearly reached full marks in the pre-test and have no room for improvement are further eliminated. Eliminated samples are moved to a waiting queue, where subsequent stages complete the responses and trigger reorganization. This ensures the stability of the peer-teaching gain under small sample conditions, reducing the risk of misjudging valid explainers.

[0022] Group control establishes a hard condition for assigning at least one effective presenter to each learning group, which is then automatically recalculated via boundary signals during inter-group intervals. In a static roster, there is no guarantee that a group will have capable and effective presenters, and this cannot be changed by grouping based on individual indices. In this invention, each group is equipped with data-verified presenters when entering the next round of peer teaching. The regularized allocation through rotation and serpentine completion further compresses the recalculation process within inter-group intervals, improving timeliness.

[0023] Limiting the range of mean participation between groups prevents highly engaged learners from clustering in a few groups, while pairing rotation prevents the same explainer-mentee relationship from repeating in two consecutive rounds. The former ensures that the active base of discussions in each group is similar, while the latter ensures that the gain samples of the explainers come from different mentors, and that mentors come into contact with more explainers within a semester. The coverage of peer teaching pairing is significantly expanded, and the distribution of mentor gain within the class converges with each round, thus suppressing the misallocation of peer teaching resources in large classes.

[0024] All data collection relies solely on key events from the response and role keys. After alignment with the boundary signals, each person's and each step's interactive behavior is instantly recorded within the classroom. Compared to manual recall and recording after class, the completeness and timeliness of the recording are significantly improved. Compared to solutions that introduce video or voice analysis, it eliminates the expense of imaging equipment and the burden of processing personal image data, allowing for the modification and deployment of ordinary multimedia classrooms.

[0025] For students whose participation falls below the lower quartile of their class, the system immediately issues a downgraded task package, allowing them to resume answering easier questions and increase evaluable events. For students whose participation has exceeded the upper quartile but has not yet been verified, the system pushes a prompt to apply for a lecture role, guiding potential lecturers into the role. Incentive indicators are only activated in tiers when the target is met for the first time during the semester, and are linked to teacher prompts. Feedback occurs at the moment of behavioral change, simultaneously suppressing the motivation for repeated rewards and score-chasing, thus improving the timeliness and directionality of classroom feedback loops.

[0026] The classroom wireless environment experiences brief interruptions, resulting in a lack of historical data available for the first lesson of the semester, and a temporary shortage of presenters. Local caching at student stations, combined with recalculation triggered by communication recovery and aligned to process boundaries, ensures that events during the interruption are not lost and their definitions remain unchanged. Recalculation does not backtrack to previously executed groups, avoiding secondary disruptions to classroom order. Groups with missing presenters retain the student configuration from the previous round and are manually assigned by the teacher. The first lesson starts with a cold start based on pre-test scores, ensuring the control loop maintains operation under these conditions. All process data, along with values ​​before and after recalculation, are appended to the student's progress record with a version number, allowing authorized read-only access on the supervisor's evaluation terminal. This transforms previously scattered paper records into a traceable chain of process evidence, enhancing the reliability of anomaly handling and the auditability of process data. Attached Figure Description

[0027] Figure 1 This is a system structure block diagram of Embodiment 1 of the present invention; Figure 2 This is a flowchart of the group control process in Embodiment 2 of the present invention; Figure 3 This is a flowchart illustrating the anomaly recovery process in Embodiment 3 of the present invention; Figure 4 This is a flowchart of the comparative experiment in Embodiment 4 of the present invention. Detailed Implementation

[0028] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] Example 1 like Figure 1As shown, this embodiment uses a large-class classroom of "Primary School Mathematics Curriculum and Teaching" for primary education majors at a certain university as the implementation environment. The class size is 48 students, the course lasts 16 weeks per semester, and each class is set up with 5 classroom segments according to the lesson script: pre-class task assessment, explanation, group task, peer teaching, and peer review and summary. The segment scripts are pre-loaded on the teacher's control device. 48 desktop student stations are deployed in the classroom, numbered... to The student station is an embedded button terminal, equipped with a response key group, role keys, and indicator units. The response key group includes four option keys and one submit key. The role keys include a request for explanation key and a request for help key. The indicator unit is a combination of a three-color LED ring and a two-digit digital display. The teacher's control device uses a lectern tablet terminal. The data processing device uses an edge computing host deployed in the classroom's power distribution cabinet, configured with a quad-core processor and 8GB of memory. The group indicator device consists of eight screens suspended above the eight group seating areas. The student station, teacher's control device, and group indicator device are all connected to the data processing device via a 2.4GHz wireless LAN communication interface, with an event message timestamp accuracy of 0.1 seconds.

[0030] The student station collects two types of events. Response events are generated when a student presses the submit button, with fields including student station identifier, stage identifier, question identifier, response result, and response delay. The response delay is the time difference between the question being sent and the submission button being triggered. Role events are generated when a student presses the role button, with fields including student station identifier, stage identifier, role type, paired object, and timestamp. Role types are divided into two categories: requesting explanation and requesting help. When a student presses the request for explanation button in the peer-tutoring stage, the data processing device registers the student as an explanation student and registers the other members of their learning group as assisted students, establishing a peer-tutoring relationship between explanation students and assisted students. When a student presses the request for help button, the data processing device adds the student as an assisted student under the explanation student's name in their group. In the peer review summary stage, peer review questions are answered and submitted by the response button group, and the response event carrying the peer review question type identifier is counted as a peer review submission event. The teacher-side control device advances the class according to the lesson plan, sending a lesson boundary signal to the data processing device at the start and end times of each lesson. The lesson boundary signal carries the lesson type and start and end times.

[0031] The data processing device first preprocesses the received events: key events that fall outside the stage boundary are directly discarded; repeated triggers of the same key within a 500-millisecond window are counted only once, and this debouncing window is taken from the bouncing characteristic parameters in the key device specification sheet; the preprocessed events are written into the event buffer after being aligned with the stage identifier and timestamp.

[0032] Questions are organized into a question bank according to knowledge points. Each question carries a knowledge point identifier, a difficulty level identifier, and a time limit for the question type. The main example knowledge point in this embodiment is written addition with carrying within 100. This knowledge point includes vertical calculation questions with step-by-step scoring of the result and carry marker. The answer result... The automatic scoring system assigns four values: 0, 0.4, 0.6, and 1.0. A correct answer scores 0.6, and a correct carry-in scoring scores 0.4. This question type tests both vertical calculation and the standardized expression of carry-in marking. The supplementary example uses the perimeter of a rectangle and a square to illustrate the consistency judgment of knowledge points. (Question delay limit) Based on a pre-set question bank, 30 seconds are allotted for objective multiple-choice questions, and 120 seconds for step-by-step scoring vertical questions. These time limits are designed based on the course team's question bank's response time. Question content is presented uniformly by question set on the teaching projection screen; individualized task packages and guidance prompts are displayed on the student station's digital screen using question set numbers. Students locate the corresponding question set according to the number and answer accordingly. The response time delay begins from the moment the numbered instruction is issued. The score for a single response is... Calculated according to formula (1): in The value assigned to the response result for this response comes from the automatic scoring of the response event; The response delay for this response originates from the response event; This is the upper limit of the delay for this question, which is preset from the question bank; The time delay conversion factor is set to 0.3. This value was selected through calibration in a trial class during the first two weeks of the course, within a calibration range of 0.2 to 0.4. The calibration criterion is that the score fluctuation of the same student when repeatedly answering the same knowledge point with variations of equal difficulty does not exceed 0.1. The value is rounded to two decimal places, ranging from 0 to 1. Multiplying the result by 100 gives the corresponding percentage score, which is then used to input the pre-test score and post-test score of the event pair.

[0033] The data processing device assembles event pairs based on classroom segments. Within the segment sequence of the same lesson, for each peer-teaching relationship, the most recent response from the student before the peer-teaching relationship was established, matching the knowledge point identifier of the question in the explanation registration, is taken as the pre-test response. The response from the same student after the peer-teaching relationship was established, on a question with the same knowledge point and a variation of the same difficulty level, is taken as the post-test response. The pre-test response, peer-teaching relationship, and post-test response are assembled into an event pair. Responses from students on perimeter-related questions are not paired with the explanation registration for addition knowledge points due to inconsistent knowledge point identifiers. The assembled event pairs are then screened by a quality gate, which includes two criteria: first, a time window upper limit criterion, where event pairs with an interval exceeding 12 minutes between the establishment of the peer-teaching relationship and the post-test response are discarded. This upper limit is taken from the designed segment duration to ensure that the post-test occurs immediately after the explanation; second, a lower limit criterion for the number of student samples, where the number of student samples falls below the lower limit. The explanation and registration process still calculates the mutual teaching gain as usual, but the gain for this round is recorded as low confidence. A value of 3 is chosen, based on the small sample statistical stability baseline. Based on the above criteria, this embodiment applies the pre-test response results... Event pairs that already have a perfect score of 1.0 and a previous test score of at least 0.90 will be removed due to insufficient room for improvement. Event pairs removed by the time window or the perfect score criteria will be moved to the waiting queue and reassembled after the same student completes their answers in subsequent rounds.

[0034] For event pairs that pass through the quality gate, the data processing device calculates the results according to formula (2). Mutual teaching gains : in To explain to students A group of students who have passed the quality gate under my name. The number of students in this set is derived from the mutual learning relationship registration; For the students receiving assistance The pre-test score, For the students receiving assistance The post-test scores are all calculated using formula (1); The range of values ​​is to The result, rounded to two decimal places, is used for valid instructor assessment and development records. When the value is not lower than 3, the vessel The case of high confidence is denoted as high confidence, and all other cases are denoted as low confidence. The rules for degrading the low confidence gain are explained in Example 3. (When explaining to students...) High confidence mutual teaching gain Reaching the gain threshold At that time, the student giving the explanation is marked as a valid explainer; The smaller value between the upper quartile of the peer teaching gain distribution for the current class and the preset baseline value of 0.05 is selected and updated continuously as the class progresses. This distribution uses the peer teaching gains of all students giving lectures in the current class as samples. When this smaller value is less than zero... Take zero, and set the baseline value to 0.05. Convert the baseline value to 5 points according to the aforementioned percentage system. The value-added evaluation baseline is set by the course group.

[0035] Student participation According to formula (3): in This represents the number of steps already completed in this lesson, derived from the step boundary signal count; The sequence number of the classroom segment; The four types of interactive events are: submitting a response, requesting an explanation, submitting a post-test and peer review after requesting help. For students In the Within the first classroom segment The count of class events comes from the event cache; The weights for the scale are 1.0, 2.0, 1.0 and 1.0 for the four categories respectively, which are taken from the preset weights of the course quality evaluation scale; the cap value of 5 in the formula is the upper limit of the sum of weighted event scores for each student in each stage, which is taken from the scale and is used to suppress repeated key presses to farm scores. The value is converted to a percentage based on this formula, rounded to one decimal place, and then sent to the group balancing assessment, task classification, and growth record.

[0036] The boundary signal at the end of the peer teaching session triggers the generation of group instructions. The data processing device divides the 48 trainees into eight learning groups of six, with each group constrained to have at least one effective presenter. Effective presenters are rotated into the eight groups in descending order of peer teaching gain, while the remaining trainees fill the remaining groups based on their participation. Detailed rules for inter-group balancing and pairing rotation are elaborated in Example 2. The group instruction fields include the group number, a list of member trainee station identifiers, and the presenter identifier within the group. The data processing device outputs the group instructions to eight group screens and each trainee station indicator unit. The group screens display the group member numbers and presenter numbers within the group. The trainee station indicator units display the target group number on a digital screen and use light ring colors to distinguish whether a trainee is a presenter. Trainees adjust their seating area according to the indicator units, and group changes are completed within the intervals between sessions. Besides the group screen format, the group instruction device can also be implemented via the teacher's control device's projection interface; changes in implementation do not affect the generation and execution logic of the group instructions.

[0037] Taking the second lesson of week 3 as an example, an end-to-end execution chain is given. In the pre-task assessment, the whole class completed a step-by-step, graded vertical calculation problem on the knowledge point of addition with carrying within 100; in the peer teaching session, students stood... The student registered as a lecturer by clicking the "Apply for Lecturer" button, and 5 members in their group were registered as assisted students. The pre-test response result was 1.0 and the pre-test score was 0.90. Based on the full score elimination clause, the student was removed from the group, and the group of students receiving assistance was... , , and There are 4 people in total, which meets the requirements. The pre-test scores and post-test scores of the four individuals are converted according to formula (1) as follows: The pre-test response result was 0.6, the response delay was 68.5 seconds, and the score was 0.50. The post-test response result was 1.0, the response delay was 52.3 seconds, and the score was 0.87. The pre-test response result was 0.4, the response delay was 95.2 seconds, and the score was 0.30; the post-test response result was 0.6, the response delay was 71.0 seconds, and the score was 0.49. The student's pre-test response result was 1.0, response delay was 48.6 seconds, and score was 0.88. The post-test response result was 1.0, response delay was 76.4 seconds, and score was 0.81. The student's score dropped due to the longer response delay in the post-test. The pre-test response result was 0.4, the response delay was 102.7 seconds, and the score was 0.30. The post-test response result was 1.0, the response delay was 63.8 seconds, and the score was 0.84. Substituting these values ​​into formula (2) yields... When the upper quartile of the inter-class teaching gain distribution for that lesson was 0.14, , For high confidence and reaching the threshold, Those marked as valid presenters; the boundary signal at the end of this stage triggers the generation of group instructions. There are a total of 9 valid presenters in this round. The first 8 are rotated into the eighth group in descending order of mutual teaching gain, and the 9th presenter is transferred to the completion sequence sorted by participation. They were assigned to Group 3, and Group 3 screens displayed that they were the presenters in the group. The target group number was displayed synchronously on the indicator units of each presenter's station. The whole class completed the group switching and entered the peer review and summary stage during the break between sessions.

[0038] The working process of this embodiment is as follows: The teacher-side control device sends out the boundary signal of the segment according to the segment script. The student stands in the segment and continuously collects response events and role events and uploads them through the communication interface. The data processing device completes preprocessing, event pair assembly and quality gate screening. The score, mutual teaching gain and participation are calculated according to formula (1) to formula (3). During the segment gap, the boundary signal triggers the generation of group instructions according to the group constraints. The group instructions are executed by the group screen and indicator unit. The segment rolls to form a closed loop of collection, processing, control and re-collection. All events and calculation results are written into the growth file of the student in each segment. The communication interruption of the student station is triggered by no response for 3 consecutive heartbeat cycles for a total of 15 seconds. The recovery condition is that the heartbeat message arrives again. The priority of the buffer back-transmission calculation after recovery is lower than that of the current round of group instruction generation. The complete rollback process of this abnormal chain is carried out in embodiment 3.

[0039] The change in scores of assisted learners on variations of the same knowledge point is the most direct and measurable output of the effect of the teaching behavior on the learning outcome. Attributing this change to the teaching learners makes the effectiveness of mutual teaching, which could only be judged by the after-class impression, a recalculated control quantity. The response result and response delay are combined and converted by formula (1), so that the score reflects both correctness and proficiency. Even if the assisted learners answer correctly and answer faster after the teaching, it will be reflected as an improvement in score, thus expanding the discriminative power of the gain. The upper limit of the time window in the quality gate ensures that the post-test occurs immediately after the teaching, eliminating the confounding effects of other teaching activities. The lower limit of the sample size reduces the small sample gain. By suppressing low-confidence participants from being marked as effective instructors, and minimizing misjudgments caused by occasional fluctuations in individual learners, while eliminating perfect scores avoids diluting the gain of samples with no room for improvement, these three factors collectively ensure the stability of the peer teaching gain. Group constraints evenly distribute effective instructors across learning groups, ensuring that each group has data-verified instructional supply in the next round of peer teaching. Combined with rolling recalculation triggered by boundary signals, grouping is adjusted step-by-step based on classroom performance, synchronizing with the classroom pace. Weighted scores capped for each stage and scale weights prevent participation from inflating due to repeated keystrokes, and the percentage-based caliber provides a unified benchmark for subsequent task grading and balance determination. The aforementioned data collection structure, gate caliber, attribution calculation, and group constraints are interdependent, jointly supporting individual-by-individual record-keeping and dynamic allocation of peer teaching resources in large-class classrooms.

[0040] Example 2 like Figure 1 and Figure 2 As shown, this embodiment is a detailed implementation of group control and incentive feedback based on embodiment 1. The implementation environment, device deployment, event acquisition, and calculation methods for score conversion, mutual teaching gain, and participation are all the same as in embodiment 1. The time delay conversion factor is different. Take 0.3 as the lower limit of the sample size of the assisted students. 3, upper limit of time window is 12 minutes, scale weights The values ​​1.0, 2.0, 1.0, and 1.0 are all taken from the values ​​and sources in Example 1. The response tasks in the explanation phase are randomly assigned by the data processing device according to participation levels. The assignment rule is to include students whose current participation level is lower than the class median in the priority assignment pool, so that low-participation students have the opportunity to respond during the phase. This rule serves to cover the participation level data collection scope without changing the statistical caliber of participation.

[0041] When the grouping instruction is generated, the data processing device adds two constraints to the grouping constraints described in Example 1. One of them is the inter-group balance constraint, which is determined according to formula (4): in and For any two study groups and The mean participation rate within the group is calculated from the participation rates of the 6 members of the group. The arithmetic mean is obtained. Calculated according to the participation statistics formula described in Example 1; To balance the threshold, a score of 10 out of 100 is used, derived from the acceptable baseline for inter-group differences in the course quality evaluation scale. A true result allows the grouping scheme to proceed, while a false result triggers an exchange adjustment. Secondly, there is a pairing rotation constraint. The data processing device maintains a pairing record table. The key field of the table is a combination of the instructor's identifier and the assisted student's identifier, and the value field is the grouping round in which the pairing was most recently established and the cumulative number of pairings. After each round of grouping instructions is issued, the registration results are updated according to the mutual teaching relationship. During the allocation process, if assigning a valid instructor to a group would cause a pairing in the table to be established repeatedly in two adjacent rounds, then that instructor will be moved one position up the rotation sequence and assigned to the next group.

[0042] The allocation process is as follows: when the effective presenters in a round are sorted from high to low according to mutual teaching gain, they are rotated into eight groups, one presenter per group. When the number of presenters exceeds eight, the extra presenters are transferred to the normal supplementary sequence. The remaining students are sorted from high to low according to participation and are added to each group in a serpentine order until each group has 6 students. Then, the formula (4) is executed to determine the outcome. If the determination fails, the non-presenter with the highest participation in the group with the highest average participation is selected and swapped with the non-presenter with the lowest participation in the group with the lowest average participation. The determination is re-evaluated and the upper limit of the iteration of the swap adjustment is 3 times. This upper limit is taken from the engineering constraint of the recalculation time of the interval between links. If the upper limit is reached and the result still fails, the grouping instruction is issued according to the scheme with the smallest current range, and the balance failure prompt is pushed to the teacher's control device.

[0043] Tasks are tiered and executed upon arrival of the boundary signal at each stage. The data processing device continuously calculates the lower quartile values ​​based on the class participation distribution of the stages already completed in the current lesson. and the upper quartile value When student participation is lower than At that time, a reduced-level task package is issued to the student's learning station. The task package is object-oriented data with fields including task package identifier, difficulty level, list of question set identifiers, recipient learning station identifier, and issuance time. The question set of the reduced-level task package consists of variations of questions at a lower difficulty level within the same knowledge point. The difficulty level is determined by the pre-set difficulty level identifiers in the question bank. For the main example knowledge point, the set of questions consists of vertical addition problems within 100 without carrying. When the student's participation rate is not lower than If the student is not marked as a valid presenter, a prompt to apply for a presenter guide will be issued to the student station. The digital screen of the indicator unit will display the guide prompt code and the light ring will flash slowly to guide the student to switch to the presenter role in the peer teaching session.

[0044] The incentive feedback is gating through three types of achievement events: the first is when the peer teaching gain of the instructors first reaches the gain threshold. The second category is when student participation surpasses the upper quartile of the class for the first time. The third category is when a student's post-test score is higher than their pre-test score for the first time. For all three categories, the "first time" criterion is the first time within the course semester. The determination status is recorded in the student's growth portfolio using three Boolean fields and updated progressively. When any type of achievement event occurs, the data processing device sends an incentive indicator change instruction to the corresponding student station's instruction unit. The incentive indicator has three levels: achieving one type of event activates the first level, achieving two types upgrades to the second level, and achieving three types upgrades to the third level. The indicator lights are displayed in white, blue, and gold, respectively. Simultaneously, a prompt message is pushed to the teacher's control device, with fields including the student station identifier, achievement event type, and achievement time, allowing the teacher to provide immediate verbal affirmation.

[0045] Taking the end of the peer teaching session in the first lesson of week 5 as an example, the average participation rates of the eight groups after the initial allocation were 28.4, 33.6, 27.1, 30.2, 26.5, 22.2, 29.3, and 31.0 respectively, with a maximum range of 11.4. This triggers a swap adjustment, selecting non-lecturing students with a participation rate of 41.5 from the second group with the highest average participation. Compared to the participants in the 6th group with the lowest average participation rate of 19.0, this was different from the participants in the 6th group. After swapping, the means of the two groups become 29.9 and 26.0 respectively, and the range of the eight groups decreases to 5.0. Formula (4) is passed, one iteration is completed, and the grouping instruction is issued; at the same boundary time, the distribution of class participation is... 20.8 It is 37.5. Participation rate 19.0% is lower than The student station received a reduced-level task package consisting of vertical addition problems within 100 without carrying. The participation rate of 39.6% is higher than Those who fail to provide effective explanations will receive a prompt to apply for explanation guidance at their student station; assisted students For the first time this semester, if a post-test score is higher than a pre-test score, the indicator unit will light up a white incentive icon, and the teacher will receive a notification of achievement simultaneously.

[0046] Regarding the operation process of a lesson, in the closed loop of data acquisition and calculation in Example 1, the boundary signal of each stage triggers a round of task classification and incentive gating judgment. The boundary signal at the end of the peer teaching stage additionally triggers group recalculation with double additional constraints. The pairing record table is updated with each round of grouping. The data processing device takes the stage event flow as input and outputs the three types of instructions of grouping, task and incentive through the communication interface to the instruction unit and group screen for execution. The teacher receives two types of prompts: balance and achievement.

[0047] Intergroup equilibrium constraints prevent group fragmentation caused by the clustering of highly engaged students, ensuring that discussions in each group are based on similar levels of activity and that the supply of explanations and the demand for assistance are roughly matched between groups. Pairing and rotation constraints prevent the continuous repetition of the same pair of explainers and students, expanding the coverage of each explainer's assistance and ensuring that the peer-teaching gain samples come from different students, thus better reflecting the explanation ability itself. On the other hand, it prevents students from becoming dependent on a single explainer. The exchange adjustment involves swapping the extreme members of the two end groups, which is the step that maximizes the contribution to the range while keeping the other six groups unchanged. Combined with the iteration upper limit, it ensures that recalculation is completed within the inter-stage intervals. The reduced-order task package allows low-engagement students to recover their success rate on easier questions first, increasing the number of evaluable events from the collection side. The guidance prompts introduce highly engaged students who have not yet verified their explanation abilities into the explanation role, expanding the pool of effective explanation candidates. Both of these factors make it easier to satisfy the grouping constraints in subsequent rounds. The incentive indicators are lit in tiers based on the first achievement of the target within the semester, focusing feedback on the moment of behavioral change rather than repeated rewards, suppressing the motivation to rack up points while retaining the information content of the instructions.

[0048] Example 3 like Figures 1 to 3 As shown, this embodiment expands upon Embodiments 1 and 2 by implementing confidence downgrading, anomaly recovery, and growth profiles. The implementation environment and device deployment are the same as in Embodiment 1. The calculation methods for score conversion, peer teaching gain, participation, and inter-group balance determination, as well as the lower limit of the sample size of assisted learners, are also specified. The values ​​and sources for setting 3 and the upper limit of the time window to 12 minutes are the same as those in Examples 1 and 2.

[0049] After the confidence marker is determined according to the terms described in Example 1, the low-confidence mutual teaching gain is downgraded according to the following rules: low-confidence gain does not participate in the valid presenter marking and therefore does not enter the group constraint determination; low-confidence gain reaches the gain threshold. The incentive flag change of the first type of achievement event in Example 2 is still triggered, so that the explanation behavior under small sample conditions can be fed back without affecting the group decision; the low confidence gain and its confidence flag are written into the growth file together, and recalculated after the subsequent assisted samples of the same explanation student are supplemented. The recalculation result is added to the new version of the current round record in the form of version number increment and the confidence is re-determined.

[0050] The student station and the data processing device exchange heartbeat messages every 5 seconds. If the data processing device does not receive a response from a student station for 3 consecutive heartbeat cycles, it determines that the communication of that student station is interrupted, with an interruption determination delay of 15 seconds. During the interruption, the response event and role event of that student station are written to the station's local storage module. The cache capacity of the local storage module is designed with a margin of twice the peak value of events in a single class, taking 2000 events. After communication is restored, the student station sends back the cached events with the original timestamp. The data processing device aligns the returned events to their respective stages according to the start and end times of each stage recorded by the stage boundary signal, reassembles the event pairs, and recalculates the mutual teaching gain of the relevant instructors. The recalculation is performed during the stage gap and has a lower priority than the generation of the group instruction in the current round. The recalculation result only updates the growth file and the input of the next round of grouping. Group instructions that have already been issued and executed are not retroactively canceled.

[0051] When the number of highly confident, effective presenters is less than 8 times the number of learning groups, the vacant group retains the presenter configuration from the previous round of grouping. This grouping is recorded in the student profile with a reuse marker. Simultaneously, the data processing device sends a manual designation prompt to the teacher's control device. The prompt fields include the vacant group number and a list of candidate students. Candidate students are ranked in descending order of mutual learning gain. After the teacher selects a candidate, the designation result is written to the current round's grouping instruction and student profile with a manual designation marker. A cold start is performed for the first lesson when there is no historical data: the first round of grouping uses a descending serpentine grouping based on the pre-test scores from the pre-task assessment. The initial participation value is assigned based on the percentage value of the pre-test score, derived from the cold start clause of the course quality evaluation scale. From the second stage onwards, grouping is switched to be driven by mutual learning gain and participation.

[0052] The growth portfolio is stored in the file storage module of the data processing device, with the student as the primary key and accumulated by class. The fields of each stage record are student station identifier, class number, stage identifier, participation, peer teaching gain, confidence flag, event pair detail reference, three types of achievement event status, task package record, reuse or manually specified flag, record version number and update time. Each stage boundary signal triggers a write operation, and the update triggered by the supplementary calculation appends the version number in an incrementing manner without overwriting the original record. The supervision and evaluation terminal can access the file in read-only mode after authorization by the teacher's control device. Access can be retrieved by class, student and class. The stage records of the 16 weeks of the whole semester constitute the data foundation for the student's growth evaluation and the data source for the evaluation process.

[0053] Taking the first lesson of week 9 as an example, during the group task session, students stood... After three consecutive heartbeat cycles without response, communication was deemed interrupted. During the interruption, 41 events were buffered within the station. Communication was restored and transmitted back during the peer teaching phase. The data processing device aligned the transmitted events to the group task phase and reassembled them. As the event pair of the assisted student passed the quality gate, the score change of the event pair was 0.19, and the students in its group who gave the presentation... The peer teaching gain was adjusted from 0.11 for 3 assisted learners to 0.13 for 4 assisted learners. The adjusted result was written into the file and entered into the next round of group input. The groups already executed in the current round remained unchanged. At the end of the peer teaching session of the same class, there were a total of 6 high-confidence effective presenters in the current round, which was less than 8 groups. Groups 4 and 7 continued to use the presenter configuration of the previous round and triggered manual designation prompts. The teacher designated the presenters from the candidate list on the teacher's end. For the 7th group of trainees, the instructions are to write the records into the file as instructed by the group.

[0054] The operation process under abnormal conditions is as follows: heartbeat detection continuously monitors the communication status of each student station. Stations that are interrupted are transferred to local cache, and after recovery, they are sent back for supplementary calculation. Supplementary calculation is aligned along the boundary of the process and does not backtrack to the already executed groups. When there is a shortage of lecturers, the group constraints can be maintained by using two paths with manual designation. Cold start is initiated by the pretest score to start the first round of grouping. All events, calculations, tags and designated records are written to the growth file step by step according to versioned fields and provided to the supervisor evaluation terminal for read-only access.

[0055] Heartbeat criteria and local caching ensure that the data acquisition chain does not lose events during short-term interruptions in the classroom wireless environment. After transmission, the data is recalculated according to the link boundary, ensuring that the event caliber of the mutual teaching gain is consistent before and after the anomaly. The recalculation does not backtrack to the already executed grouping because the classroom grouping action is physically irreversible. Backtracking would only introduce secondary disturbances, while the recalculation results can be imported into the next round of input, allowing the control closed loop to correct itself within one link cycle. Low confidence gain is isolated from grouping decision but connected to incentive feedback, so that small sample noise does not drive incorrect grouping adjustments, and does not discourage the enthusiasm of the teacher under low sample conditions. When there is a shortage of teachers, the use of the previous round of configuration is the strategy with the least disturbance to the grouping state. The manual path designation incorporates the teacher's judgment into the closed loop and leaves a trace. Both of these together ensure that the constraint of the teacher configuration for each group can still be maintained when there is insufficient data. Versioned append writing of the archives makes the values ​​before and after the recalculation traceable, providing an uncovered process evidence chain for supervision and evaluation.

[0056] Example 4 like Figures 1 to 4As shown, this embodiment presents a comparative experiment. The experiment was conducted in three parallel classes of the same year in the primary education major at the same undergraduate institution, with 48 students in each class. All students used the same classroom script, the same question bank, and the same course quality evaluation scale parameters from "Primary School Mathematics Curriculum and Teaching". The semester lasted 16 weeks, with two classes per week. The three classes were taught by three teachers with similar teaching experience from the same course group, and they were rotated once at the end of the 8th week according to a predetermined plan to balance the teacher factors. The pre-test and post-test for the three classes used the same question bank with variations of the same knowledge points and difficulty. The scoring method was consistent with the scoring conversion method described in Embodiment 1. The three classes differed only in their classroom interaction recording and group control methods: Control class A followed the existing practices, with fixed groups determined at the beginning of the semester. Classroom interaction performance was manually recalled and recorded by teachers after class, and pre-tests and post-tests were organized and entered using a unified testing method; Control class B deployed the same student station data collection hardware as in Example 1, with grouping balanced between groups by the system based on individual student answer accuracy and answer activity, and mutual teaching association registration, mutual teaching gain calculation, and group lecturer configuration constraints were disabled; Experimental class C operated with the complete configuration of Examples 1 to 3. All indicators follow the standards disclosed in Examples 1 to 3: the completeness rate of individual interaction event records is calculated as the ratio of the number of entries recorded by the system or manually to the sampling verification benchmark; the time taken for a single round of group adjustment is calculated from the generation of the grouping instruction to the issuance of the group change instruction; for Class A, the time taken for one manual rearrangement is calculated; the achievement rate of each group's configured data-verified instructors is calculated according to the effective instructor standard of Experimental Class C; the post-test gain of assisted students is calculated according to the score change of the mutual teaching gain described in Example 1, with its class mean and standard deviation; the mutual teaching pairing coverage rate is calculated as the ratio of the number of different instructors actually contacted by each student during the semester to the total number of instructors. The mutual teaching relationship, instructor identity, and pre- and post-test pairing of Control Classes A and B are recorded by the in-class observers according to a unified record sheet, and the corresponding indicators are calculated offline using the same score conversion and gain standard as Experimental Class C. The summary results for 16 weeks are shown in Table 1.

[0057] Table 1. Comparison Results of Three Parallel Classes over 16 Weeks Under the three-class comparison conditions, the completeness rate of individual interaction event recording in experimental class C was comparable to that in control class B, and both were significantly higher than that in control class A, which used manual recording. This indicates that the improvement in recording completeness comes from the student station collection structure itself. The time taken for single-round group adjustment in experimental class C was higher than that in control class B. The extra 2.5 seconds came from the computational cost of group constraint judgment and exchange adjustment iteration. The upper bound of this cost was locked by the iteration limit of 3 times, which was still completed within the inter-stage gap. Regarding the achievement rate of each group's configured instructors, control class B only had students in the group who reached the gain threshold under accidental circumstances. Experimental class C achieved almost full achievement due to group constraints and the missing downgrade path. The rounds that were not achieved corresponded to the rounds used in Example 3. The mean post-test gain of assisted students in experimental class C was higher than that in control class B, and the class standard deviation was significantly lower than that of the two control classes. The distribution of gain within the class tended to converge with pairing rotation. The mutual teaching pairing coverage rate in experimental class C was significantly higher than that in the two control classes, corresponding to the direct effect of pairing rotation constraints. Control group B and experimental group C share the same data acquisition hardware, but differ only in the attribution and group constraint stages. The mechanism of this setting is to locate the differences in the indicators of the two classes in Table 1 at the mutual teaching gain attribution and the instructor configuration constraints themselves.

[0058] The three parallel classes operated in parallel for 16 weeks: Experimental class C followed the complete closed loop of Examples 1 to 3, with student station events flowing through the communication interface to the data processing device, and abnormal rounds being handled by following Example 3 and manually specified backtracking paths; Control class B performed balanced allocation of individual indicators on the same data acquisition chain; Control class A performed manual recording and fixed grouping; all indicators were recorded weekly and summarized at the end of the term as Table 1.

[0059] The above description is merely a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be modified within the scope of the concept described herein by means of the above teachings or the technology or knowledge in related fields.

Claims

1. A smart teaching data acquisition and processing intelligent control system, characterized in that, The system includes at least two student stations, a teacher-side control device, a data processing device, and a grouping indicator device. The student stations, teacher-side control device, and grouping indicator device are connected to the data processing device via communication interfaces. Each student station has a response key group, a role key, and an indicator unit. The response key group is used to collect student response events during class sessions. Each response event includes a student station identifier, session identifier, question identifier, response result, and response delay. The role key is used to collect role events, including requests for explanation and requests for help. Requests for explanation events register the mutual teaching relationship between the explaining student and the assisted student. The teacher-side control device sends session boundary signals to the data processing device. These signals mark the type and start / end times of each class session. The data processing device receives the response events, role events, and session boundary signals. Within the same class session sequence, it uses the response events of assisted students on questions belonging to the same knowledge point before the mutual teaching relationship is established as pre-test responses and then uses them in the subsequent mutual teaching sessions. The response events after the establishment of the teaching relationship are used as post-test responses, assembled into event pairs with corresponding role events, and quality gate screening is performed on the event pairs. The quality gate screening includes at least determining the lower limit of the number of assisted learner samples and the upper limit of the time window from the establishment of the teaching relationship to the post-test response. The data processing device is also used to attribute the change in the assisted learner's post-test response score relative to the pre-test response score to the corresponding instructor for the event pairs that pass the quality gate screening, statistically obtain the instructor's teaching gain, and mark the instructor as a valid instructor when the instructor's teaching gain reaches the gain threshold. The data processing device is also used to generate grouping instructions based on grouping constraints triggered by the segment boundary signal during class breaks. The grouping constraints include at least one valid instructor in each learning group. The data processing device outputs the grouping instructions to the grouping indicator device and the indicator unit of the learner station. The grouping indicator device and the indicator unit of the learner station execute the grouping instructions, instructing learners to complete the learning group adjustment according to the grouping instructions.

2. The system according to claim 1, characterized in that, The data processing device is also used to perform weighted statistics on the number of interactive events of each student in the already conducted class sessions according to the preset scale weights, so as to obtain the student's participation level. The interactive events include at least response events, request for explanation events, and peer review submission events.

3. The system according to claim 1, characterized in that, The gain threshold is the smaller of the upper quartile of the peer teaching gain distribution of the class in the current lesson and the preset benchmark value. When the smaller value is less than zero, the gain threshold is zero and is updated on a rolling basis with the class.

4. The system according to claim 2, characterized in that, The grouping constraints also include that the range of the average participation rate among learning groups is not greater than a preset equilibrium threshold, and that the mutual teaching pairing of the same instructor and the same assisted student is not repeated in two adjacent rounds of grouping.

5. The system according to claim 4, characterized in that, When the data processing device generates grouping instructions, it rotates and assigns effective instructors to each learning group in descending order of mutual teaching gain, and then fills in the remaining students into each learning group in a serpentine order according to their participation.

6. The system according to claim 1, characterized in that, The data processing device binds confidence flags to mutual teaching gains. When the number of assisted learner samples is not less than the lower limit of the sample size, the confidence flag is high; otherwise, the confidence flag is low. Mutual teaching gains with low confidence flags do not participate in the determination of effective instructor labeling and grouping constraints.

7. The system according to claim 2, characterized in that, When a student's participation rate is lower than the lower quartile of the class participation rate distribution for that class, the data processing device sends a downgraded task package to the student's student station. When a student's participation rate is not lower than the upper quartile of the class participation rate distribution for that class and the student has not been marked as a valid instructor, the data processing device sends an application prompt for an instructor to the student's student station.

8. The system according to claim 2, characterized in that, When any one of the following three events occurs: the peer teaching gain of the instructor reaches the gain threshold for the first time, the student's participation level exceeds the upper quartile of the class participation distribution for the first time, or the assisted student's post-test response score is higher than the pre-test response score for the first time, the data processing device sends an incentive flag change instruction to the instruction unit of the corresponding student station and pushes a prompt message to the teacher control device.

9. The system according to claim 1, characterized in that, When communication between the student station and the data processing device is interrupted, the student station caches the response event and role event in the local storage module. After communication is restored, the response event and role event are sent back with timestamps. The data processing device recalculates the mutual teaching gain according to the step boundaries. The recalculation result is only used to update the growth profile of the student and the input of the next round of grouping. When the number of effective instructors is less than the number of learning groups, the data processing device uses the instructor configuration of the missing learning groups from the previous round of grouping and sends a manual assignment prompt to the teacher control device.

10. The system according to claim 1, characterized in that, The data processing device writes each student's participation, peer learning gains, event pair details, and confidence markers for each class session into the growth portfolio storage module. The growth portfolio is accumulated by class and can be accessed by the teacher control device and the supervision and evaluation terminal.

Images