AI enabling quantitative teaching evaluation method and system based on multi-modal data portrait

Abstract

The invention provides an AI enabling quantitative teaching evaluation method and system based on a multi-modal data portrait, and relates to the technical field of teaching evaluation. Multi-source teaching behavior data in a teaching scene are synchronously collected; performing feature extraction on the multi-source teaching behavior data according to the cognitive load label to obtain a behavior semantic unit chain of students in the teaching scene; according to the interaction description information, determining the dynamic participation density when the students receive knowledge transmission, and further determining the teaching response efficiency when the students respond to the teaching strategy in the teaching scene according to the dynamic participation density; and determining a teacher ability evaluation portrait based on teaching performance according to the behavior semantic unit chain and the teaching response efficiency, and generating a teaching evaluation report according to the teacher ability evaluation portrait. According to the invention, when the teaching evaluation result lacks strategy adaptability, the teaching behavior of the teacher can be subjected to dynamic structured association evaluation, so that the accuracy of teaching evaluation is improved.

Description

Technical Field

[0001] This application relates to the technical field of teaching evaluation. More specifically, this application relates to an AI empowered quantitative teaching evaluation method and system based on multi-modal data portraits. Background Art

[0002] Teaching evaluation refers to a whole-process evaluation activity that, in a real teaching scenario, based on multi-dimensional data such as teachers' teaching behaviors, students' learning responses, and the usage of teaching resources, measures and analyzes the teaching process and teaching effects in a systematic and structured manner to reflect the teaching level of teachers and the learning achievements of students. Traditional teaching evaluations mainly rely on single or subjective data such as questionnaires, grades, and class observation records, and it is difficult to comprehensively reflect teaching dynamics and students' cognitive changes. With the development of artificial intelligence and multi-modal perception technologies, teaching evaluation has gradually evolved towards data-driven, process-oriented, and intelligent feedback directions. By synchronously collecting multi-modal data such as audio, video, handwriting, eye movement, and interaction behaviors in teaching, and combining semantic recognition, behavior modeling, and intelligent analysis technologies, refined portraits of teachers' teaching and students' cognitive trajectories can be constructed, thereby evaluating teachers' teaching.

[0003] However, in existing AI empowered quantitative teaching evaluations based on multi-modal data portraits, there are deficiencies in aspects such as the accuracy of teaching behavior semantic modeling, the single dimension of student cognitive response analysis, and the insufficient quantification of teacher-student interaction dynamics. This makes the teaching evaluation results lack strategy adaptability, and it is difficult to truly reflect the actual impact of teachers' teaching strategies on students' learning efficacy, resulting in fragmented feedback results of teaching portraits and lacking personalized guiding significance, thus restricting the application ability of teaching evaluation in precise teaching improvement. Therefore, how to perform dynamic structured correlation modeling on teachers' teaching behaviors when the teaching evaluation results lack strategy adaptability to improve the accuracy of teaching evaluation is a problem faced by the industry. Summary of the Invention

[0004] This application provides an AI empowered quantitative teaching evaluation method and system based on multi-modal data portraits, which can perform dynamic structured correlation evaluation on teachers' teaching behaviors when the teaching evaluation results lack strategy adaptability to improve the accuracy of teaching evaluation.

[0005] In a first aspect, this application provides an AI empowered quantitative teaching evaluation method based on multi-modal data portraits. The evaluation method includes the following steps:

[0006] Synchronously collect multi-source teaching behavior data in the teaching scenario;

[0007] Determine the cognitive load labels of students' participation in the classroom in the teaching scenario, and perform feature extraction on the multi-source teaching behavior data according to the cognitive load labels to obtain the behavior semantic unit chain of students in the teaching scenario;

[0008] Obtain the interactive description information of teachers and students in the teaching scenario, determine the dynamic participation density of students when receiving knowledge transfer according to the interactive description information, and then determine the teaching response effectiveness of students in the teaching scenario when responding to teaching strategies based on the dynamic participation density;

[0009] Determine the teacher ability evaluation portrait based on teaching performance according to the behavior semantic unit chain and the teaching response effectiveness, and generate a teaching evaluation report according to the teacher ability evaluation portrait.

[0010] In this embodiment, the multi-source teaching behavior data refers to the original information set collected by different types of perception devices in the teaching site for depicting the teaching behaviors of teachers and the learning behaviors of students.

[0011] In this embodiment, determining the cognitive load label of students participating in the classroom in the teaching scenario specifically includes:

[0012] Collect the behavioral performance data of students participating in the classroom in the teaching scenario;

[0013] Determine the cognitive interaction index of the cognitive state of students according to the behavioral performance data;

[0014] Determine the cognitive load label of students participating in the classroom in the teaching scenario based on the cognitive interaction index.

[0015] In this embodiment, the behavior semantic unit chain refers to a behavior sequence composed of multiple atomic semantic units that are continuously connected in the time dimension and have a teaching semantic logical relationship.

[0016] In this embodiment, obtaining the interactive description information of teachers and students in the teaching scenario specifically includes:

[0017] Synchronously collect the multi-modal teaching interaction stream and generate an interaction event sequence of teachers and students in the teaching scenario;

[0018] Determine the teaching behavior semantics in the teaching scenario according to the interaction event sequence;

[0019] Generate the interactive description information of teachers and students in the teaching scenario according to the teaching behavior semantics.

[0020] In this embodiment, the interactive description information refers to the structured data of the interactive characteristics of teachers and students in the teaching process.

[0021] In this embodiment, the dynamic participation density refers to an evaluation index of the total amount of effective feedback made by students on the knowledge content transmitted by teachers per unit time within a teaching cycle.

[0022] In this embodiment, determining the teaching response efficacy when students respond to teaching strategies in the teaching scenario based on the dynamic participation density specifically includes:

[0023] Determining the density fluctuation characteristics of students in the teaching strategy according to the dynamic participation density;

[0024] Determining the collaborative response rule when students respond to the teaching strategy according to the density fluctuation characteristics;

[0025] Determining the teaching response efficacy when students respond to the teaching strategy in the teaching scenario according to the collaborative response rule.

[0026] In this embodiment, determining the teacher ability evaluation portrait based on teaching performance according to the behavioral semantic unit chain and the teaching response efficacy specifically includes:

[0027] Analyzing the knowledge transfer trajectory when the teaching strategy is implemented according to the behavioral semantic unit chain; determining the cognitive feedback attributes of students in the teaching scenario through the teaching response efficacy;

[0028] Determining the teacher ability evaluation portrait based on teaching performance according to the knowledge transfer trajectory and the cognitive feedback attributes.

[0029] In a second aspect, the present application provides an AI empowered quantitative teaching evaluation system based on a multi-modal data portrait, which is used to execute an AI empowered quantitative teaching evaluation method based on a multi-modal data portrait. The evaluation system includes:

[0030] A data acquisition module, which is used to synchronously collect multi-source teaching behavior data in the teaching scenario;

[0031] A feature extraction module, which is used to determine the cognitive load label of students' participation in the classroom in the teaching scenario, and perform feature extraction on the multi-source teaching behavior data according to the cognitive load label to obtain the behavioral semantic unit chain of students in the teaching scenario;

[0032] An evaluation response module, which is used to obtain the interactive description information between teachers and students in the teaching scenario, determine the dynamic participation density of students when receiving knowledge transfer according to the interactive description information, and further determine the teaching response efficacy when students respond to the teaching strategy in the teaching scenario based on the dynamic participation density;

[0033] A report generation module, which is used to determine the teacher ability evaluation portrait based on teaching performance according to the behavioral semantic unit chain and the teaching response efficacy, and generate a teaching evaluation report according to the teacher ability evaluation portrait.

[0034] The technical solutions provided by the embodiments disclosed in the present application have the following beneficial effects:

[0035] Synchronously collect multi-source teaching behavior data in the teaching scenario; determine the cognitive load labels of students' participation in the classroom in the teaching scenario, and perform feature extraction on the multi-source teaching behavior data according to the cognitive load labels to obtain the behavioral semantic unit chain of students in the teaching scenario; obtain the interactive description information of teachers and students in the teaching scenario, determine the dynamic participation density of students when receiving knowledge transfer according to the interactive description information, and then determine the teaching response efficiency of students in the teaching scenario when responding to teaching strategies according to the dynamic participation density; determine the teacher ability evaluation portrait based on teaching performance according to the behavioral semantic unit chain and the teaching response efficiency, and generate a teaching evaluation report according to the teacher ability evaluation portrait.

[0036] Therefore, it can be seen that in this application, the accuracy of teaching evaluation can be improved in the case of lack of strategy adaptability in teaching evaluation results; among them, by synchronously collecting multi-source teaching behavior data in the teaching scenario, the full-dimensional coverage and time synchronization marking of students' behavior characteristics in the teaching process are realized, ensuring the collection consistency and semantic coherence of teaching behavior data, so as to construct a unified behavior data fusion framework under the condition of multi-source heterogeneous input, effectively solving the problems of fragmented teaching behavior observation and difficult time series alignment in existing evaluations; by determining the cognitive load labels of students' participation in the classroom in the teaching scenario, the accurate perception and dynamic mapping of the implicit cognitive state in students' behavior can be realized, so as to construct a behavioral semantic unit chain model driven by cognitive load, effectively enhancing the semantic correspondence between teaching behavior and students' mental state, and improving the cognitive interpretation ability of behavior analysis results; by obtaining the interactive description information of teachers and students in the teaching scenario, the quantitative evaluation of students' response activity under strategy guidance is realized, effectively solving the problem that it is difficult to capture the dynamic fluctuation of students' participation status in existing evaluation methods; by constructing the causal association path between the teacher's knowledge transfer trajectory and the student's cognitive feedback attributes, a teacher ability evaluation portrait for the teaching process can be formed, and a personalized teaching evaluation report can be generated through the structured analysis of each ability dimension in the teacher ability evaluation portrait, effectively improving the process traceability and result applicability of the evaluation content, so as to realize the precise feedback mechanism for teaching effectiveness.

[0037] In summary, the technical solution adopted in this application can perform dynamic structured correlation evaluation on teachers' teaching behaviors when the teaching evaluation results lack strategy adaptability, so as to improve the accuracy of teaching evaluation. Brief Description of the Drawings

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0039] Figure 1 is an exemplary flowchart of an AI empowered quantitative teaching evaluation method based on multi-modal data portraits provided by this application;

[0040] Figure 2 is a schematic flowchart of determining the behavioral semantic unit chain provided by this application;

[0041] Figure 3 is a schematic flowchart of determining the dynamic participation density provided by this application;

[0042] Figure 4 is a module structure diagram of an AI empowered quantitative teaching evaluation system based on multi-modal data portraits provided by this application. Detailed implementation manners

[0043] Next, the technical solutions in the embodiments of this application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of this application.

[0044] The embodiments of this application provide an AI empowered quantitative teaching evaluation method and system based on multi-modal data portraits. The core is to synchronously collect multi-source teaching behavior data in the teaching scenario; determine the cognitive load labels of students' participation in the classroom in the teaching scenario, and perform feature extraction on the multi-source teaching behavior data according to the cognitive load labels to obtain the behavioral semantic unit chain of students in the teaching scenario; obtain the interactive description information between teachers and students in the teaching scenario, determine the dynamic participation density of students when receiving knowledge transfer according to the interactive description information, and further determine the teaching response efficiency of students in the teaching scenario when responding to teaching strategies based on the dynamic participation density; determine the teacher ability evaluation portrait based on teaching performance according to the behavioral semantic unit chain and the teaching response efficiency, and generate a teaching evaluation report according to the teacher ability evaluation portrait.

[0045] In Embodiment 1, to better understand the above technical solutions, the above technical solutions will be described in detail below in conjunction with the accompanying drawings of the specification and specific implementation manners. Refer to Figure 1 As shown, this figure is an exemplary flowchart of an AI empowered quantitative teaching evaluation method based on multi-modal data portraits shown in this embodiment of this application. The evaluation method includes the following steps:

[0046] In step S1, synchronously collect multi-source teaching behavior data in the teaching scenario.

[0047] In specific implementation, first, deploy multiple sensing terminals such as visual acquisition devices (high-definition cameras, depth cameras), audio acquisition devices (multi-channel microphone arrays), and interactive acquisition terminals (such as electronic whiteboards, student tablets, click recording systems) in a real teaching scenario. Then, each device needs to be configured with a unified time synchronization protocol (such as using the Network Time Protocol NTP service) to ensure consistent data timing. After the collected data is preliminarily processed by the edge computing module, data encoding is performed, and the data is uploaded to the teaching behavior data fusion module in a unified data format (such as a JSON structure). Among them, the data fusion module performs unified alignment of visual, auditory, and operation data from different channels in the spatial and temporal dimensions, and performs subject normalization based on person recognition (such as face recognition and ID binding). Finally, a structured teaching behavior data record form is output, and multi-source teaching behavior data can be obtained from the teaching behavior data record form. In other embodiments, other methods may also be used to determine multi-source teaching behavior data, which will not be elaborated here.

[0048] It should be noted that in this application, multi-source teaching behavior data refers to the set of original information collected from different types of sensing devices in a teaching scene and used to depict the teaching behavior of teachers and the learning behavior of students.

[0049] In step S2, determine the cognitive load label of the students' participation in the classroom in the teaching scene, and perform feature extraction on the multi-source teaching behavior data according to the cognitive load label to obtain the behavioral semantic unit chain of the students in the teaching scene.

[0050] In this embodiment, the determination of the cognitive load label of the students' participation in the classroom in the teaching scene can be implemented by the following steps:

[0051] Collect the behavioral performance data of the students' participation in the classroom in the teaching scene;

[0052] Determine the cognitive interaction index of the students' cognitive state according to the behavioral performance data;

[0053] Determine the cognitive load label of the students' participation in the classroom in the teaching scene according to the cognitive interaction index.

[0054] In specific implementation, first, multi-modal perception devices are arranged in the teaching scenario, including visual acquisition devices, audio acquisition devices, and interaction acquisition devices. Among them, the visual acquisition devices include high-definition cameras and depth cameras, which are installed in the front, left, and right sides of the classroom to capture the facial expressions, head orientations, and body postures of students in real time; the audio acquisition devices include multi-channel microphone arrays for collecting the voice response frequencies and voice features (pitch, pause, rhythm) of students; the interaction acquisition devices such as the electronic tablets used by students, classroom response systems, or smart blackboard terminals are used to record the operation behaviors of students such as clicks, answering questions, dragging, and note writing. The data collected by the multi-modal perception devices is used as the behavior performance data of students' participation in the classroom in the teaching scenario. Then, visual acquisition data can be used to extract head postures (such as continuous head turning and low head frequency) and eye movement trajectories to identify whether students are paying attention to the podium, screen, or textbook. Then, the sliding time window is used to statistically calculate the continuous time period and switching frequency of students' fixation on teaching content, and the concentration score is calculated. Then, the click path, answer correct rate, and answer time collected by the interaction device are analyzed. Among them, the inverse relationship between the answer time and the correct rate can be used to judge the level of thinking load; if a student repeatedly modifies the answer or the answer duration deviates significantly from the average value, it indicates a high cognitive consumption during the understanding process. The frequency of students' participation in voice Q&A, the length of voice content, and the number of active speeches are statistically calculated, and these data are converted into behavior initiative scores, which are used as cognitive interaction indicators. Finally, the cognitive interaction indicators of all students within a fixed time period are normalized; the processed cognitive interaction indicators are input into a classification model, and the classification model classifies the current student state according to the learned weights; the output of the classification model is three-level labels: "low load", "medium load", "high load", corresponding to three states of easy understanding, moderate challenge, and cognitive overload respectively; the labels are bound to the time period and student identity and recorded in the cognitive load log table, and the cognitive load labels of students' participation in the classroom in the teaching scenario can be obtained through the cognitive load log table.

[0055] It should be noted that in this application, the behavior performance data refers to the set of external behavior information that can be observed and quantified by students during the classroom teaching process; the cognitive interaction indicator refers to the set of characteristic indicators that depict the current knowledge processing level, cognitive load state, and learning concentration degree of students; the cognitive load label refers to the cognitive interaction classification identifier of students during classroom participation.

[0056] Preferably, in this embodiment, feature extraction is performed on the basis of the cognitive load label to obtain the behavior semantic unit chain of students in the teaching scenario. Refer to Figure 2 As shown, this figure is a schematic flowchart of determining the behavior semantic unit chain in some embodiments of this application. The determination of the behavior semantic unit chain in this embodiment can be implemented by the following steps:

[0057] In step S21, align the multi-source teaching behavior data according to the timestamps of the cognitive load tags;

[0058] In step S22, determine the dynamic behavior data segments of the students in the teaching scenario;

[0059] In step S23, perform behavior detection on the multi-source teaching behavior data aligned according to the dynamic behavior data segments and timestamps to obtain the atomic semantic units of the students in the teaching scenario; in step S24, determine the behavior semantic unit chain of the students in the teaching scenario through the atomic semantic units.

[0060] In specific implementation, first, uniformly embed millisecond-level timestamps into all multi-source teaching behavior data (including video behavior sequences, voice signal records, interaction system logs, etc.). Subsequently, use the timestamps in the student cognitive load labels as anchor points to establish an alignment window based on time periods, such as each window being 10 seconds; screen the behavior records that coincide with this period in each window, and merge the video frame data, voice segments, and interaction events into a unified data structure according to the student identifier and time dimension; for channels with different sampling frequencies (such as the camera being 25 frames per second and the voice being 1 semantic transcription per second), use interpolation mapping to align the data; finally, form a multi-source teaching behavior data segment set aligned with the unified time axis, and each multi-source teaching behavior data segment set is bound to a cognitive load label and corresponds to a teaching behavior time period. Then, on the multi-source teaching behavior data after timestamp alignment, delimit a preliminary analysis window based on the time period covered by the cognitive load label, and then apply the sliding window mechanism to extract the behavior event sequences of each channel from each time window; use the action boundary detection method based on pose estimation for the video behavior sequence to identify the significant change points between adjacent frames, such as pose changes or sitting posture adjustments; for the voice channel, extract the voice activity segment (Voice Activity Detection) as a potential speaking or listening response; for the interaction log, screen the continuous event segments composed of operation events (such as clicks, swipes, answering questions), fuse the behavior event sequences of each channel, and extract the time period with continuous behavior characteristics as the student's dynamic behavior data segment through an adaptive segmentation method based on the content change rate. Then, in the visual channel, use a pose estimation model (such as OpenPose or MediaPipe) to extract the key point coordinate sequence; judge the action type according to the movement trajectory and spatial position change of the key points in the key point coordinate sequence, such as "raising the head", "raising the hand", "lowering the head to write". Among them, use a speech recognition tool (such as Tencent Speech Transcription or Google Speech) in the voice channel to extract the speech transcription text; perform keyword matching and sentiment analysis on the semantic content to identify behaviors such as "actively answering", "talking to oneself", "repeating listening". Analyze the click path and timestamp sequence in the interaction channel to identify interaction behaviors such as "browsing content", "answering questions operation", "page switching"; distinguish "concentrated operation" and "intermittent response" through the breakpoints and operation density in the key point coordinate sequence. Encode the action combinations identified within each behavior segment into an atomic semantic unit. Each atomic semantic unit contains the action type, the executing subject, the time boundary, and the associated cognitive load label.Finally, the time-ordered atomic semantic units are input into the behavioral semantic construction module, including: clustering and summarizing multiple atomic behavior sequences using a sequence analysis method based on the hidden Markov model; assigning semantic labels to each behavior sequence pattern, such as "focused learning chain", "distraction chain", and "repeated understanding chain"; and merging atomic semantic units with continuity and semantic consistency into behavioral semantic unit chains; each behavioral semantic unit chain contains start time, end time, semantic type, corresponding cognitive load level, and triggering environment information.

[0061] It should be noted that, in this application, timestamp alignment refers to matching multi-source teaching behavior data according to a unified time axis, so that data from different acquisition channels can be jointly analyzed within the same time window; a dynamic behavior data segment refers to a teaching behavior segment that is continuous in time and coherent in behavioral performance; an atomic semantic unit refers to a specific teaching interactive action completed by students within a short period of time, such as "looking at the screen", "lowering your head to write", "raising your hand to speak", etc.; a behavior semantic unit chain refers to a behavior sequence composed of multiple atomic semantic units that are continuously connected in the time dimension and have a teaching semantic logical relationship.

[0062] In step S3, the interaction description information between teachers and students in the teaching scenario is obtained, and the dynamic participation density of students when receiving knowledge transfer is determined based on the interaction description information, and then the teaching response effectiveness when students respond to teaching strategies in the teaching scenario is determined by the dynamic participation density.

[0063] In this embodiment, obtaining the interaction description information between teachers and students in the teaching scene can be achieved by using the following steps:

[0064] Synchronously collect multimodal teaching interaction flows and generate interaction event sequences between teachers and students in teaching scenarios;

[0065] Determining the teaching behavior semantics in the teaching scenario according to the interaction event sequence;

[0066] Generate interaction description information between teachers and students in the teaching scene based on the teaching behavior semantics.

[0067] In specific implementation, first, during synchronous acquisition, cameras are deployed in the teaching environment to capture the limb behaviors and spatial position changes of teachers and students; microphone arrays are installed to obtain voice audio streams for identifying speakers and semantic content; teaching system logs are accessed, such as click-on-answer, raise-hand button, and electronic blackboard operation records; if there are touch devices, touch trajectories and interaction gestures are accessed; a unified timestamp in milliseconds is added to all acquisition channels; data segments with timestamps in different channels are window-level matched (e.g., the sliding window is 1 second); within the same time window, they are merged into an interaction event record according to the event priority and primary-secondary order; then an interaction event sequence is constructed in chronological order, such as "teacher's explanation → student nods → student answers questions → teacher's feedback". Then, the interaction event sequence is used as input, and local event segments are extracted in a sliding window manner; a sequence labeling model based on conditional random fields is applied to semantically label the events in the local event segments, such as "questioning behavior segment", "answer response segment", "error correction feedback segment"; the labeling results are classified to identify teaching behavior semantic tags, and context information is added to the identified behavior semantics, such as "current knowledge point", "interaction location (beginning, middle, end)", "teacher-student interaction times"; a teaching behavior semantic structure including semantic categories, behavior initiators, behavior response chains, and teaching environment tags is formed, and this teaching behavior semantic structure is used as the teaching behavior semantics in the teaching scenario. Finally, the teaching behavior semantics within a certain time period are summarized, such as in segments of 5 minutes; then the occurrence frequencies of various types of interaction behaviors are counted (such as the number of times teachers ask questions and the number of times students actively respond); and the interaction directions (initiated by teachers, initiated by students), whether there is a closed loop (whether there is response feedback), and the duration of each behavior semantics are marked. Next, interaction density indicators are introduced: the number of effective interaction behaviors occurring per unit time; interaction integrity indicators are introduced: whether the effective interaction chain includes start, response, and feedback; interaction diversity indicators are introduced: the coverage of interaction behavior types (such as questions, answers, comments, collaboration), and the introduced indicators are structured to generate interaction description information. For example: Based on the interaction description information, interaction description statements can be automatically written through rule templates or natural language generation models. In this class, the teacher initiated 16 knowledge point questions, 87% of which received effective responses from students; the students initiated 4 active inquiries, the classroom interaction density is relatively high, and the feedback closed-loop rate reached 92%.

[0068] It should be noted that in this application, the multi-modal teaching interaction flow refers to a continuous information flow that expresses the interaction behavior between teachers and students and is jointly composed of multiple information carriers (such as images, voices, click logs, touch behaviors, screen projection data, etc.) during the teaching process; the interaction event sequence refers to a time series set formed by recording, sorting, and classifying various teaching-significant interaction behaviors included in the multi-modal teaching interaction flow in event units; the teaching behavior semantics refers to the interaction semantic units of teaching intentions and teaching process nodes; the interaction description information refers to the structured data of the interaction characteristics between teachers and students during the teaching process.

[0069] Preferably, in this embodiment, the dynamic participation density of students during knowledge transfer is determined according to the interaction description information, with reference to Figure 3 As shown, this figure is a schematic flowchart for determining the dynamic participation density in some embodiments of this application. The dynamic participation density in this embodiment can be implemented by the following steps:

[0070] In step S31, response time markers in the teaching scenario are extracted from the interaction description information;

[0071] In step S32, the dynamic participation interval of students during knowledge transfer is determined;

[0072] In step S33, an active response index of students during knowledge transfer is determined according to the response time marker and the dynamic participation interval;

[0073] In step S34, the dynamic participation density of students during knowledge transfer is determined according to the active response index.

[0074] In specific implementation, first, identify "knowledge transfer behaviors" in the interactive description information, including teaching, example illustration, task assignment, etc.; for each knowledge transfer event initiated by the teacher, extract its start time as the "transfer starting point"; calculate whether there are interactive behaviors such as students' voice responses, answering behaviors, eye gaze, note-taking, etc. within a limited time window (e.g., within 0 to 15 seconds) after the knowledge transfer event; label the events with valid responses with the "valid response" label, and record the response delay time to form a response timeliness mark. Then, take each teaching behavior of the teacher during the teaching process as an anchor behavior node and label it as a "knowledge input event"; according to the interaction model, set the valid response interval for each input event, for example, extend it backward by 10 to 30 seconds as the dynamic participation interval. Screen all students' behavior streams according to the time stamp, only retain the behavior events that appear within each dynamic participation interval, and bind them to the corresponding knowledge input events. After the binding is completed, obtain the dynamic participation interval of the students when receiving knowledge transfer. Then, extract the students' behavior events within the dynamic participation interval and classify them into language types (answering, questioning), action types (raising hands, nodding), operation types (answering questions, clicking), and perception types (eye following, taking notes) according to the behavior types; assign weights to each type of response, for example, getting 2 points for actively answering questions and 1 point for nodding; combine the response delay time to construct a delay correction factor, for example, multiplying the score for a response within 3 seconds by 1.2 and multiplying the score outside 10 seconds by 0.8; calculate the "response score" corresponding to each knowledge transfer event; and map each response behavior of each student into an active response index. Finally, divide the entire teaching process into several time periods (e.g., one unit every 5 minutes); within each time period, count the total score of the students' active response indexes; calculate the response score density value of this time period, that is, the score divided by the time length; and conduct a standard deviation analysis on the response score density values of all the students in the class to establish an overall benchmark; normalize the response score density value of each student with the class average level to obtain the dynamic participation density.

[0075] It should be noted that in this application, the response timeliness mark is the time delay characteristic information indicating the response of students to the teacher's knowledge transfer behavior during a specific period in the teaching process; the dynamic participation interval is the time range of the response window that extends backward with the teacher's knowledge transfer behavior as the anchor on the classroom time axis; the active response index is the quality evaluation measure for measuring whether students can produce meaningful outputs after cognitive input; the dynamic participation density is the evaluation index of the total amount of effective feedback made by students on the knowledge content transferred by the teacher per unit time within a teaching cycle.

[0076] In this embodiment, the teaching response efficacy when students respond to the teaching strategy in the teaching scenario determined by the dynamic participation density can be realized by the following steps:

[0077] Determine the density fluctuation characteristics of students in the teaching strategy according to the dynamic participation density;

[0078] Determine the collaborative response rule when students respond to the teaching strategy according to the density fluctuation characteristics;

[0079] Determine the teaching response efficacy when students respond to the teaching strategy in the teaching scenario according to the collaborative response rule.

[0080] When specifically implemented, first, preset teaching strategy tags in the teaching process, such as lectures, interactive Q&A, group discussions, scenario simulations, etc., and mark time tags for each segment of strategy activities on the time axis. Map the dynamic participation density to the time period corresponding to each teaching strategy to form a sequence of "strategy type - time segment - density value". Plot this sequence of "strategy type - time segment - density value" as a curve, and obtain the maximum value of the slope in this curve as the density fluctuation characteristics of students in the teaching strategy. Then, divide the students into multiple analysis sub - groups according to optional dimensions such as academic performance level, personality preference, or historical participation behavior model; perform time alignment and normalization processing on the density fluctuation characteristics of all students in each sub - group under the same teaching strategy, so that the participation density can be compared under the same standard dimension; use the Dynamic Time Warping method or the multi - dimensional Euclidean distance calculation method to evaluate the similarity degree between the fluctuation curves; determine whether there is an obvious group response pattern of participation peak synchronization, trough synchronization, or reverse trend; if there is an obvious group response pattern of participation peak synchronization, trough synchronization, or reverse trend, then use this pattern as the collaborative response rule, such as "The scenario simulation strategy is likely to trigger an overall participation peak, but the duration is short", etc. Finally, map all the teaching strategies in the teaching scenario to the corresponding collaborative response rules of the students one by one to construct a data structure of "strategy name - response rule - student group - density response". Set three response efficacy dimensions: strong response (high density, consistent behavior), medium response (medium fluctuation, partially consistent), weak response (low density or scattered behavior); assign scores to each teaching strategy in each group according to the matching degree between the actual fluctuation characteristics and the collaborative rules; conduct a comprehensive evaluation of the overall response of the teaching strategy to obtain the teaching response efficacy of this teaching strategy in this teaching scenario.

[0081] It should be noted that in this application, the density fluctuation characteristics refer to the change trend and fluctuation degree of the dynamic participation density over time under different teaching strategy interventions; the collaborative response rule refers to an index indicating whether students generate a consistent response pattern under specific teaching strategy interventions; the teaching response efficacy refers to an index of the participation quality, reaction density, and behavior coherence shown by students in the teaching process in response to specific teaching strategies imposed by teachers.

[0082] In step S4, based on the behavioral semantic unit chain and the teaching response effectiveness, a teacher ability evaluation portrait based on teaching performance is determined, and a teaching evaluation report is generated according to the teacher ability evaluation portrait.

[0083] In this embodiment, the determination of the teacher ability evaluation portrait based on teaching performance according to the behavioral semantic unit chain and the teaching response effectiveness can be implemented by the following steps:

[0084] Analyze the knowledge transfer trajectory when the teaching strategy is implemented according to the behavioral semantic unit chain; determine the cognitive feedback attributes of students in the teaching scenario through the teaching response effectiveness.

[0085] Determine the teacher ability evaluation portrait based on teaching performance according to the knowledge transfer trajectory and the cognitive feedback attributes.

[0086] Specifically, when implementing, first, based on the student's behavioral semantic unit chain, extract the behavioral segments that are stimulated by the teacher's behavior and significantly responded by the student, such as raising hands after asking questions, taking notes after explanations, operating after demonstrations, etc.; map each explicit behavioral segment back to the corresponding teacher teaching strategy type, such as lecturing, guiding, discussing, heuristic questioning, etc. Use the teaching behavior coding table for rule matching; label the name and category of the knowledge points transmitted for each teaching strategy node, such as concept type, principle type, skill type, problem-solving type, etc., to form teaching content units; then connect the teaching strategy nodes and student behavior response nodes in sequence according to the time stamp to construct a "strategy behavior - response behavior - knowledge point" path sequence to obtain the knowledge transfer trajectory. Then, extract corresponding indicators such as participation density, behavior delay, and interaction depth from the teaching response effectiveness; if the density is high, the response is rapid, and the behavior is rich, it is marked as "high cognitive feedback"; if the density is medium and the response is intermittent, it is marked as "medium cognitive feedback"; if there is no response or the lag is obvious, it is marked as "low cognitive feedback"; summarize the feedback annotations of students during the transfer of different knowledge points, and statistically calculate the high, medium, and low feedback ratios according to the knowledge point dimension to form the cognitive feedback attributes of students in the teaching scenario. Finally, set the evaluation dimensions of the teacher ability evaluation portrait, including teaching strategy selection ability: the ability to match appropriate teaching strategies for different knowledge points; knowledge organization and expression ability: the coherence and logic of the knowledge path; student cognitive stimulation ability: the proportion of promoting high cognitive feedback; teaching regulation ability: the frequency and effectiveness of strategy adjustment in the face of low cognitive feedback. Then use a weighted scoring mechanism to score the teaching strategy selection ability, knowledge organization and expression ability, student cognitive stimulation ability, and teaching regulation ability item by item; at the same time, construct a visualized portrait atlas, including a radar chart, a strategy distribution chart, a feedback heat map, etc.; form a three-layer mapping model of "behavior path - student feedback - teacher ability" and output the final teacher ability evaluation portrait.

[0087] It should be noted that in this application, the knowledge transfer trajectory refers to the behavioral path by which teachers promote the gradual transfer of knowledge to students' cognitive structures through specific teaching behaviors in classroom teaching; the cognitive feedback attribute refers to the reaction characteristics of students' cognitive understanding, knowledge absorption, and behavioral expression under the stimulation of specific teaching strategies and knowledge content; the teacher ability evaluation portrait refers to a comprehensive evaluation model that reflects the teacher's ability levels in dimensions such as strategy application, teaching control, and cognitive stimulation, extracted through the overall analysis of the knowledge transfer path and the quality of students' responses in the teaching process.

[0088] In addition, in specific implementation, generating a teaching evaluation report based on the teacher ability evaluation portrait can be achieved in the following manner: First, standardize the data of each ability dimension in the teacher ability evaluation portrait and conduct a horizontal comparison with historical teaching performance data to identify the performance strengths and improvement areas of the teacher in dimensions such as teaching strategy application, knowledge transfer logic, and cognitive stimulation efficiency. Subsequently, based on a predefined multi-dimensional teaching evaluation report template, dynamically embed contents such as ability indicators, students' feedback, typical teaching segments, and strategy adaptability into the report structure to automatically generate a teaching evaluation report combining text and graphics.

[0089] It should be noted that in this application, the teaching evaluation report refers to a comprehensive document that presents the teaching behavior performance, students' cognitive feedback, and the analysis results of teaching effectiveness of teachers within a specific teaching cycle.

[0090] Thus, it can be seen that in this application, the accuracy of teaching evaluation can be improved in the case where the teaching evaluation results lack strategy adaptability; among them, by synchronously collecting multi-source teaching behavior data in the teaching scenario, achieving full-dimensional coverage and time-synchronized marking of students' behavioral characteristics in the teaching process, ensuring the consistency and semantic coherence of the collection of teaching behavior data, thereby constructing a unified behavioral data fusion framework under multi-source heterogeneous input conditions, effectively solving the problems of fragmented observation of teaching behaviors and difficult temporal alignment in existing evaluations; by determining the cognitive load labels of students' participation in the classroom in the teaching scenario, the accurate perception and dynamic mapping of the implicit cognitive state in students' behaviors can be achieved, thereby constructing a cognitive load-driven behavioral semantic unit chain model, effectively enhancing the semantic correspondence between teaching behaviors and students' mental states, and improving the cognitive interpretation ability of behavioral analysis results; by obtaining the interactive description information of teachers and students in the teaching scenario, the quantitative evaluation of students' response activity under strategy guidance can be achieved, effectively solving the problem that it is difficult to capture the dynamic fluctuations of students' participation status in existing evaluation methods; by constructing a causal association path between the teacher's knowledge transfer trajectory and the students' cognitive feedback attributes, a teacher ability evaluation portrait for the teaching process can be formed, and a personalized teaching evaluation report can be generated through the structured analysis of each ability dimension in the teacher ability evaluation portrait, effectively improving the process traceability and result applicability of the evaluation content, thereby realizing a precise feedback mechanism for teaching effectiveness.

[0091] In summary, the technical solution adopted in this application can perform dynamic structured correlation evaluation on teachers' teaching behaviors when the teaching evaluation results lack strategy adaptability, so as to improve the accuracy of teaching evaluation.

[0092] Embodiment 2. This application provides an AI empowered quantitative teaching evaluation system based on multi-modal data portraits. Refer to Figure 4 As shown, this figure is a module structure diagram of an AI empowered quantitative teaching evaluation system based on multi-modal data portraits according to this embodiment of this application. The evaluation system includes:

[0093] A data acquisition module 100, configured to synchronously acquire multi-source teaching behavior data in a teaching scenario;

[0094] A feature extraction module 200, configured to determine the cognitive load labels of students' participation in the classroom in the teaching scenario, and perform feature extraction on the multi-source teaching behavior data according to the cognitive load labels to obtain a behavior semantic unit chain of students in the teaching scenario;

[0095] An evaluation response module 300, configured to obtain the interactive description information of teachers and students in the teaching scenario, determine the dynamic participation density of students when receiving knowledge transfer according to the interactive description information, and further determine the teaching response efficiency of students in the teaching scenario when responding to teaching strategies from the dynamic participation density;

[0096] A report generation module 400, configured to determine a teacher ability evaluation portrait based on teaching performance according to the behavior semantic unit chain and the teaching response efficiency, and generate a teaching evaluation report according to the teacher ability evaluation portrait.

[0097] This application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of this application. It should be understood that each process and / or block in the flowcharts and / or block diagrams can be implemented by computer program instructions, and the combination of processes and / or blocks in the flowcharts and / or block diagrams can also be implemented by computer program instructions. These computer program instructions can be provided to the processors of general-purpose computers, special-purpose computers, embedded processors, or other programmable data processing devices to generate a machine, so that the instructions executed by the processors of the computer or other programmable data processing devices generate a device for implementing the specified functions in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0098] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program, and this program can be stored in a computer-readable storage medium. The storage medium includes read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc memories, magnetic disc memories, tape memories, or any other medium that can be used to carry or store data and is computer-readable.

[0099] It should also be noted that the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent in such a process, method, commodity or device. Without more limitations, an element defined by the statement "including one..." does not exclude the existence of another identical element in the process, method, commodity or device including the element.

Claims

1. An AI empowerment quantitative teaching evaluation method based on multi-modal data portraits, characterized in that, The evaluation method includes the following steps: Synchronously collect multi-source teaching behavior data in the teaching scenario; Determine the cognitive load labels of students' participation in the classroom in the teaching scenario, and extract features from the multi-source teaching behavior data according to the cognitive load labels to obtain the behavior semantic unit chain of students in the teaching scenario; Obtain the interactive description information of teachers and students in the teaching scenario, determine the dynamic participation density of students when receiving knowledge transfer according to the interactive description information, and then determine the teaching response efficacy of students in the teaching scenario when responding to teaching strategies based on the dynamic participation density; Determine the teacher ability evaluation portrait based on teaching performance according to the behavior semantic unit chain and the teaching response efficacy, and generate a teaching evaluation report according to the teacher ability evaluation portrait.

2. The AI energy-giving quantitative teaching evaluation method based on multimodal data portraits according to claim 1, characterized in that, The multi-source teaching behavior data refers to the original information set collected from different types of perception devices in the teaching site and used to depict teachers' teaching behaviors and students' learning behaviors.

3. The AI empowerment quantitative teaching evaluation method based on multimodal data portraits as claimed in claim 1, wherein Determining the cognitive load labels of students' participation in the classroom in the teaching scenario specifically includes: Collect the behavior performance data of students' participation in the classroom in the teaching scenario; Determine the cognitive interaction index of the students' cognitive state according to the behavior performance data; Determine the cognitive load labels of students' participation in the classroom in the teaching scenario based on the cognitive interaction index.

4. The AI empowerment quantitative teaching evaluation method based on multi-modal data portraits as claimed in claim 1, wherein, The behavior semantic unit chain refers to a behavior sequence composed of multiple atomic semantic units that are continuously connected in the time dimension and have a teaching semantic logical relationship.

5. The AI empowerment quantitative teaching evaluation method based on multimodal data portraits according to claim 1, wherein Obtaining the interactive description information of teachers and students in the teaching scenario specifically includes: Synchronously collect multi-modal teaching interaction flows and generate an interaction event sequence of teachers and students in the teaching scenario; Determine the teaching behavior semantics in the teaching scenario according to the interaction event sequence; Generate the interactive description information of teachers and students in the teaching scenario according to the teaching behavior semantics.

6. The AI empowered quantitative teaching evaluation method based on multimodal data portraits according to claim 1, wherein The interactive description information refers to the structured data of the interactive characteristics of teachers and students in the teaching process.

7. The AI empowerment quantitative teaching evaluation method based on multi-modal data portraits according to claim 1, characterized in that The dynamic participation density refers to an evaluation index of the total amount of effective feedback given by students to the knowledge content transmitted by teachers per unit time within a teaching cycle.

8. The AI empowerment quantitative teaching evaluation method based on multi-modal data portraits according to claim 1, characterized in that, Determining the teaching response efficacy of students in the teaching scenario when responding to teaching strategies based on the dynamic participation density specifically includes: Determine the density fluctuation characteristics of students in the teaching strategy according to the dynamic participation density; Determine the collaborative response rules of students when responding to teaching strategies according to the density fluctuation characteristics; Determine the teaching response efficacy of students in the teaching scenario when responding to teaching strategies according to the collaborative response rules.

9. The AI empowerment quantitative teaching evaluation method based on multimodal data portraits according to claim 1, characterized in that, Determining the teacher ability evaluation portrait based on teaching performance according to the behavior semantic unit chain and the teaching response efficacy specifically includes: Analyze the knowledge transfer trajectory when implementing the teaching strategy according to the behavior semantic unit chain; determine the cognitive feedback attributes of students in the teaching scenario through the teaching response efficacy; Determine the teacher ability evaluation portrait based on teaching performance according to the knowledge transfer trajectory and the cognitive feedback attributes.

10. An AI empowered quantitative teaching evaluation system based on multi-modal data portraits, which is used to execute an AI empowered quantitative teaching evaluation method according to any one of claims 1 to 9, and is characterized in that, The evaluation system includes: A data collection module for synchronously collecting multi-source teaching behavior data in the teaching scenario; A feature extraction module, which is used to determine the cognitive load labels of students' participation in the classroom in the teaching scenario, and extract features from the multi-source teaching behavior data according to the cognitive load labels to obtain the behavior semantic unit chain of students in the teaching scenario; An evaluation response module, which is used to obtain the interactive description information between teachers and students in the teaching scenario, determine the dynamic participation density of students when receiving knowledge transfer according to the interactive description information, and further determine the teaching response efficiency of students in the teaching scenario when responding to teaching strategies based on the dynamic participation density; A report generation module, which is used to determine the teacher ability evaluation portrait based on teaching performance according to the behavior semantic unit chain and the teaching response efficiency, and generate a teaching evaluation report according to the teacher ability evaluation portrait.

Images