Classroom teaching ability assessment method, device, equipment, storage medium and product

Abstract

The invention discloses a classroom teaching ability assessment method, device and equipment, a storage medium and a product, and relates to the technical field of artificial intelligence, and the method comprises the steps: carrying out the analysis of a classroom teaching video of a to-be-assessed teacher according to a multi-modal large language model, and obtaining a multi-modal analysis result; constructing a classroom teaching knowledge graph according to a multi-modal analysis result; and comparing the classroom teaching knowledge graph with a preset expert knowledge graph, and evaluating the classroom teaching ability of the to-be-evaluated teacher according to a comparison result to obtain an evaluation result. A section of classroom video is taken as a research object, classroom speech and behavior data are analyzed through a multi-modal large model, a multi-modal analysis result is extracted, a classroom teaching knowledge graph is constructed according to the multi-modal analysis result, the classroom teaching knowledge graph is compared with a knowledge graph of an expert classroom, a teacher ability assessment model is established, and a teacher ability assessment result is obtained. Therefore, teaching videos can be analyzed more deeply, teaching quality can be evaluated more accurately, and improved teaching ability suggestions can be provided.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method, apparatus, device, storage medium and product for evaluating classroom teaching ability. Background Art

[0002] Existing classroom evaluation schemes rely on manual observation, expert evaluations require significant manpower, and evaluation criteria are highly subjective, lacking fine-grained quantitative indicators. Existing intelligent education analytics products only analyze surface data (such as speaking rate and interaction frequency), lack a deep understanding of the structured logic of the teaching knowledge system, and lack the ability to dynamically model the knowledge transfer pathways during the teaching process. Existing education knowledge graphs are mostly static subject knowledge bases that are not dynamically linked to teachers' teaching behaviors. Therefore, how to conduct in-depth analysis of teaching quality and improve the accuracy of evaluation results is an urgent problem that needs to be solved. Summary of the Invention

[0003] The main purpose of this application is to provide a classroom teaching ability assessment method, device, equipment, storage medium and product, aiming to solve the technical problem that existing teaching ability assessment schemes are unable to conduct in-depth analysis of teaching quality.

[0004] To achieve the above objectives, this application proposes a method for evaluating classroom teaching ability, which includes:

[0005] Analyze the classroom teaching videos of the teachers to be evaluated based on the multimodal large language model to obtain multimodal analysis results;

[0006] Constructing a classroom teaching knowledge graph based on the multimodal analysis results;

[0007] Comparing the classroom teaching knowledge graph with the preset expert knowledge graph to obtain a comparison result;

[0008] The classroom teaching ability of the teacher to be evaluated is evaluated according to the comparison result to obtain an evaluation result.

[0009] In one embodiment, the step of parsing the classroom teaching video of the teacher to be evaluated based on the multimodal large language model to obtain the multimodal parsing result includes:

[0010] Conduct semantic analysis on the classroom speech data in the classroom teaching videos of the teachers to be evaluated to obtain classroom knowledge points;

[0011] Detecting the teacher, students, and blackboard area to be evaluated based on the classroom teaching video to obtain teaching behaviors, wherein the teaching behaviors include at least one of classroom blackboard writing, teacher behavior, teacher expression, student behavior, and student expression;

[0012] Multimodal feature alignment is performed on the classroom knowledge points and the teaching behaviors according to a multimodal large language model to obtain a multimodal analysis result.

[0013] In one embodiment, the step of constructing a classroom teaching knowledge graph based on the multimodal analysis results includes:

[0014] Identifying knowledge point entities and teaching behavior entities according to the multimodal analysis results;

[0015] Extracting logical relationships between multiple knowledge point entities;

[0016] Determining a knowledge point importance channel of the knowledge point entity according to a global context;

[0017] Determine the behavior validity channel of the teaching behavior entity according to the behavior duration;

[0018] The knowledge point importance channel and the behavior effectiveness channel are fused through a dual-channel graph attention network to obtain a joint influence weight;

[0019] Determining the teaching behavior association between the knowledge point entity and the teaching behavior entity according to the joint influence weight;

[0020] The knowledge point entity, the teaching behavior entity, the teaching behavior association and the logical relationship are weightedly associated according to the joint influence weight to obtain a classroom teaching knowledge graph.

[0021] In one embodiment, the step of weightedly associating the knowledge point entity, the teaching behavior entity, the teaching behavior association, and the logical relationship according to the joint influence weight to obtain a classroom teaching knowledge graph includes:

[0022] Constructing a node according to the knowledge point entity and the teaching behavior entity;

[0023] Establishing edges according to the logical relationship, the teaching behavior association and the spatiotemporal proximity relationship between the nodes;

[0024] Dynamic joint temporal modeling is performed based on the nodes and the edges through a temporal graph convolutional network and a long short-term memory network to generate a classroom teaching knowledge graph.

[0025] In one embodiment, the step of comparing the classroom teaching knowledge graph with the preset expert knowledge graph to obtain a comparison result includes:

[0026] Determining the similarity distance based on the shortest graph distance between nodes in the classroom teaching knowledge graph and the preset expert knowledge graph;

[0027] Extracting paths with a frequency greater than a preset frequency in the classroom teaching knowledge graph according to a random walk algorithm to obtain high-frequency paths;

[0028] Determining the path difference between the high-frequency path and the key path library in the preset expert knowledge graph;

[0029] Determine multi-dimensional ability evaluation indicators based on the classroom teaching knowledge graph and the preset expert knowledge graph;

[0030] A comparison result is generated according to the similarity distance, the path difference and the multi-dimensional capability evaluation index.

[0031] In one embodiment, the multi-dimensional ability assessment indicator includes at least one of knowledge coverage, logical coherence, behavioral effectiveness, temporal and spatial rationality, and teacher-student interaction quality;

[0032] The step of determining the multi-dimensional ability evaluation indicators based on the classroom teaching knowledge graph and the preset expert knowledge graph includes:

[0033] Determining knowledge coverage based on the number of nodes in the classroom teaching knowledge graph and the number of nodes in the preset expert knowledge graph;

[0034] Determine the logical coherence according to the number of cross-level jumps in the classroom teaching knowledge graph, the total number of knowledge points in the classroom teaching knowledge graph, and the jump frequency benchmark value in the preset expert knowledge graph;

[0035] Determining the effectiveness of the behavior based on the probability distribution of the teaching behavior association in the classroom teaching knowledge graph and the preset probability distribution in the preset expert knowledge graph;

[0036] Determine spatiotemporal rationality based on the spatiotemporal distribution of nodes in the classroom teaching knowledge graph and the spatiotemporal distribution of nodes in the preset expert knowledge graph;

[0037] The quality of teacher-student interaction is determined based on the total number of interactive behaviors initiated by the teacher in the classroom teaching knowledge graph, the student participation rate of each interactive behavior, the duration of interaction initiation, and the preset importance in the preset expert knowledge graph.

[0038] In addition, to achieve the above-mentioned purpose, the present application also proposes a classroom teaching ability evaluation device, which includes:

[0039] The multimodal parsing module is used to parse the classroom teaching videos of the teachers to be evaluated based on the multimodal large language model to obtain multimodal parsing results;

[0040] A knowledge graph construction module, used to construct a classroom teaching knowledge graph based on the multimodal analysis results;

[0041] A knowledge graph comparison module is used to compare the classroom teaching knowledge graph with the preset expert knowledge graph to obtain a comparison result;

[0042] The teaching ability evaluation module is used to evaluate the classroom teaching ability of the teacher to be evaluated based on the comparison result to obtain an evaluation result.

[0043] In addition, to achieve the above-mentioned purpose, the present application also proposes a classroom teaching ability assessment device, which includes: a memory, a processor, and a computer program stored on the memory and runnable on the processor, and the computer program is configured to implement the steps of the classroom teaching ability assessment method described above.

[0044] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium. A computer program is stored on the storage medium, and when the computer program is executed by the processor, the steps of the classroom teaching ability evaluation method described above are implemented.

[0045] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the classroom teaching ability assessment method as described above.

[0046] This application provides a method for evaluating classroom teaching ability, which obtains multimodal analysis results by parsing the classroom teaching video of the teacher to be evaluated based on a multimodal large language model; constructing a classroom teaching knowledge graph based on the multimodal analysis results; comparing the classroom teaching knowledge graph with the preset expert knowledge graph to obtain a comparison result; and evaluating the classroom teaching ability of the teacher to be evaluated based on the comparison result to obtain an evaluation result. This application takes a classroom video as the research object, analyzes classroom speech and behavior data through a multimodal large model, extracts multimodal analysis results, and constructs a classroom teaching knowledge graph based on the multimodal analysis results, compares it with the knowledge graph of the expert classroom, and establishes a teacher ability evaluation model, so as to be able to analyze the teaching video more deeply, evaluate the teaching quality more accurately, and put forward suggestions for improving teaching ability. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0048] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0049] Figure 1 A flowchart of the first embodiment of the classroom teaching ability evaluation method provided in this application;

[0050] Figure 2 A flowchart of the second embodiment of the classroom teaching ability evaluation method provided in this application;

[0051] Figure 3 A flowchart of the third embodiment of the classroom teaching ability evaluation method of this application is provided;

[0052] Figure 4 This is a schematic diagram of the module structure of the classroom teaching ability evaluation device according to an embodiment of the present application;

[0053] Figure 5 This is a schematic diagram of the device structure of the hardware operating environment involved in the classroom teaching ability evaluation method in the embodiment of this application.

[0054] The purpose, features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0055] It should be understood that the specific embodiments described herein are merely used to explain the technical solutions of the present application and are not intended to limit the present application.

[0056] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0057] The main solution of the embodiment of the present application is: to analyze the classroom teaching video of the teacher to be evaluated based on the multimodal large language model to obtain a multimodal analysis result; to construct a classroom teaching knowledge graph based on the multimodal analysis result; to compare the classroom teaching knowledge graph with the preset expert knowledge graph to obtain a comparison result; and to evaluate the classroom teaching ability of the teacher to be evaluated based on the comparison result to obtain an evaluation result.

[0058] Because existing classroom evaluation schemes rely on manual observation, expert evaluations require a significant amount of manpower, evaluation criteria are highly subjective, and lack fine-grained quantitative indicators. Existing intelligent education analytics products only analyze surface data (such as speaking speed and interaction frequency), lack a deep understanding of the structured logic of the teaching knowledge system, and lack the ability to dynamically model the knowledge transfer paths during the teaching process. Existing education knowledge graphs are mostly static subject knowledge bases that are not dynamically linked to teachers' teaching behaviors. Therefore, how to conduct in-depth analysis of teaching quality and improve the accuracy of evaluation results is an urgent problem that needs to be solved.

[0059] This application provides a solution, which is to analyze the classroom teaching video of the teacher to be evaluated according to a multimodal large language model to obtain a multimodal analysis result; construct a classroom teaching knowledge map based on the multimodal analysis result; compare the classroom teaching knowledge map with the preset expert knowledge map to obtain a comparison result; evaluate the classroom teaching ability of the teacher to be evaluated based on the comparison result to obtain an evaluation result. This application takes a classroom video as the research object, analyzes the classroom speech and behavior data through a multimodal large model, extracts the multimodal analysis results, and constructs a classroom teaching knowledge map based on the multimodal analysis results. It compares it with the knowledge map of the expert classroom and establishes a teacher ability evaluation model, so as to be able to analyze the teaching video more deeply, evaluate the teaching quality more accurately, and put forward suggestions for improving teaching ability.

[0060] It should be noted that the execution entity of the method of this embodiment can be a computing service device with classroom teaching ability assessment, network communication, and program execution functions, such as a tablet computer, personal computer, mobile phone, etc.; it can also be a classroom teaching ability assessment device with the same or similar functions. This embodiment and the following embodiments will be described using the classroom teaching ability assessment device as an example.

[0061] Based on this, the embodiment of the present application provides a method for evaluating classroom teaching ability, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the classroom teaching ability assessment method of this application.

[0062] In this embodiment, the classroom teaching ability assessment method includes steps S10 to S40:

[0063] Step S10: Analyze the classroom teaching video of the teacher to be evaluated based on the multimodal large language model to obtain a multimodal analysis result.

[0064] It is understandable that this embodiment introduces multimodal feature alignment technology, using a multimodal large language model (MLLM) to extract knowledge points, teaching behaviors, and spatiotemporal features from the classroom teaching videos of the teacher to be evaluated, thereby forming a multimodal analysis result. The MLLM is a model that combines the reasoning capabilities of a large language model (LLM) with the ability to receive, understand, and output information from multiple modalities.

[0065] In a feasible implementation, step S10 may include steps S101 to S103:

[0066] Step S101 : performing semantic analysis on classroom speech data in the classroom teaching video of the teacher to be evaluated to obtain classroom knowledge points.

[0067] It should be noted that classroom speech data from classroom instructional videos is processed here. The input audio data stream undergoes noise reduction, echo cancellation, and voice activity detection to remove silence and background noise. Features such as MFCC are extracted to provide standardized input for subsequent speech analysis. Automatic speech recognition technologies (e.g., Wav2Vec 2.0, Whisper, or Conformer) are used to transcribe the speech data into text. Speech sentiment analysis (Bi-LSTM + Attention or WavLM + Transformer) is also performed to extract the teacher's tone and emotional characteristics. Based on the transcribed text, a large model (DeepSeek) is then used to perform semantic analysis and understanding, extracting key knowledge points and keywords. Question type classification (e.g., open-ended, fill-in-the-blank, multiple-choice) and dependency parsing are performed to assess the logic of the teacher's explanation and the students' language expression ability. Furthermore, based on the students' responses, a large language model is used to assess the completeness and accuracy of the responses, providing a basis for personalized learning analysis.

[0068] Step S102 , detecting the teacher to be evaluated, the students and the blackboard area according to the classroom teaching video to obtain teaching behaviors, wherein the teaching behaviors include at least one of classroom blackboard writing, teacher behavior, teacher expression, student behavior and student expression.

[0069] It should be understood that classroom teaching videos are analyzed and detected here. Classroom teaching videos are captured using high-definition cameras. Target detection (YOLOv8, Faster R-CNN, and other models) is used to identify teachers, students, and blackboard areas. Blackboard writing in the blackboard area is also identified. Human pose estimation (OpenPose, HRNet, and other models) is used to analyze teacher gestures and student behavior. Facial Expression Recognition is combined to analyze the concentration and emotional state of teachers and students. Classroom behavior detection is performed on video data to analyze the teacher's standing, walking, and gesture interactions, and students' hand-raising, head-lowering, and distracted behaviors. Combined with teacher-student interaction analysis, a classroom interaction network is constructed based on a graph neural network (GNN). The frequency and pattern of interactions are evaluated, which together constitute the final teaching behavior.

[0070] Step S103: performing multimodal feature alignment on the classroom knowledge points and the teaching behavior according to the multimodal large language model to obtain a multimodal analysis result.

[0071] It is understandable that multimodal feature alignment technology is used to synchronize audio, video, and OCR text data (text data includes pre-acquired lesson plans, classroom blackboard notes, etc.) through time series matching. The attention mechanism is used to optimize the information fusion of different modalities to ensure data consistency in the temporal dimension. Subsequently, the Transformer is used for cross-modal feature fusion, and the LSTM or TCN model is used for time series modeling to predict classroom behavior trends and analyze changes in student participation. Multimodal analysis results are obtained to provide data support for dynamic classroom adjustments.

[0072] Step S20: construct a classroom teaching knowledge graph based on the multimodal analysis results.

[0073] It should be understood that this step is to construct a dynamic knowledge graph. Based on the multimodal analysis results, the teaching logic chain and behavioral data are integrated to generate a classroom teaching knowledge graph with a timestamp.

[0074] Step S30: compare the classroom teaching knowledge graph with the preset expert knowledge graph to obtain a comparison result.

[0075] It should be noted that classroom data corresponding to special-grade teachers is collected in advance, and an expert knowledge graph is constructed according to the above-mentioned dynamic teaching knowledge graph construction module method. The knowledge graph template of the expert-level classroom is stored, and the classroom data of special-grade teachers is collected. The ideal teaching path is marked by educational experts. This may include but is not limited to the following: ideal knowledge point coverage (for example, junior high school mathematics requires 100% coverage of Chapter 3 of the textbook), optimal teaching path (for example, the success rate increases by 23% by explaining examples first and then deriving formulas), and common misunderstandings (for example, 80% of students easily confuse the concepts of "mass" and "weight").

[0076] It is understandable that the classroom teaching knowledge graph and the preset expert knowledge graph can be compared based on graph similarity algorithms and other methods to obtain multi-dimensional and quantifiable comparison results.

[0077] Step S40: Evaluate the classroom teaching ability of the teacher to be evaluated according to the comparison result to obtain an evaluation result.

[0078] It's understandable that large models (e.g., DeepSeek) can be used based on the comparison results to generate interpretable natural language improvement suggestions. The effectiveness of the suggestions after the teacher adopts them can then be fed back into the expert graph optimization. For example, "We detected that you didn't associate the concept of 'concentration gradient' with the 'chemical reaction rate' when explaining it (the weight of this connection in the expert graph is 0.92). We recommend adding a comparative experimental demonstration."

[0079] This embodiment provides a method for evaluating classroom teaching ability. The method includes parsing a classroom teaching video of a teacher to be evaluated using a multimodal large language model to obtain a multimodal parsing result; constructing a classroom teaching knowledge graph based on the multimodal parsing result; comparing the classroom teaching knowledge graph with a preset expert knowledge graph to obtain a comparison result; and evaluating the classroom teaching ability of the teacher to be evaluated based on the comparison result to obtain an evaluation result. This embodiment uses a classroom video as the research object, parses classroom speech and behavior data using a multimodal large model, extracts a multimodal parsing result, and constructs a classroom teaching knowledge graph based on the multimodal parsing result. This graph is compared with the knowledge graph of the expert classroom to establish a teacher ability evaluation model, thereby enabling a more in-depth analysis of the teaching video, a more accurate assessment of the teaching quality, and recommendations for improving teaching ability.

[0080] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above embodiment 1 can be referred to the above introduction and will not be described in detail later. Figure 2 , step S20, the classroom teaching ability evaluation method further includes steps S201 to S207:

[0081] Step S201: identifying knowledge point entities and teaching behavior entities according to the multimodal analysis results.

[0082] It is understandable that when recognizing the multimodal analysis results, the data output by the multimodal classroom data analysis layer is used as input, including: voice text (content knowledge features, emotional labels, etc.), teacher behavior sequence (gestures, movement trajectories, interactive behaviors, etc.), blackboard / PPT / teaching plan content (OCR text, structural hierarchy, etc.), student reaction data (behavioral actions, facial expression analysis, etc.), and other contents.

[0083] It should be understood that a spatiotemporal coding mechanism is also proposed here to vectorize the appearance time and duration of knowledge points (such as t-SNE dimensionality reduction visualization). The voice text, behavioral events, and blackboard page turning actions are synchronized and aligned to a unified time axis (the error can be set to <0.5s). Then, spatial coding is performed to grid the classroom space (for example, assuming the classroom is 10m×8m long and wide, the podium is divided into 6 areas). The teacher's position coordinates are mapped to the area code. The specific area code code is as follows:

[0084] def position_encoder(x,y):

[0085] return f"Area_{int(x / / 2)}_{int(y / / 2)}"#Generate 2m×2m grid code;

[0086] Then, the spatiotemporal feature vector is encoded, and the Transformer position encoding method is used to merge the timestamp t and the spatial code s into the spatiotemporal feature vector:

[0087] PE(t,s)=[sin(t / 10000^{2i / d}),cos(t / 10000^{2i / d})||OneHot(s)],

[0088] Among them, || represents vector concatenation and d is the dimension.

[0089] It is worth noting that knowledge point extraction and relationship extraction are performed based on the above encoding. First, entity recognition is performed, using BIO annotation to identify knowledge point entities (such as "Pythagorean Theorem") and teaching behavior entities (such as "asking questions"). A teaching content recognition model, such as one fine-tuned based on BERT-Teacher, is used to identify knowledge point entities (such as "Pythagorean Theorem") and teaching behavior entities (such as "asking questions").

[0090] Step S202: extracting the logical relationship between the plurality of knowledge point entities.

[0091] It is understandable that the logical relationships between multiple knowledge point entities (such as "pre-dependency" and "comparison and analysis") are extracted, and then a triple (teacher, lecture, knowledge point) is constructed based on the logical relationship, knowledge point entity, and teaching behavior entity. The output of the entity recognition model is as follows:

[0092] {"text":"Now let's look at the solution to Example 3",

[0093] "entities":[{"start":8,"end":10,"type":"knowledge point","value":"Example 3"}]}.

[0094] It should be understood that relationship extraction is also required here, and a graph convolutional network (GCN) is used for semantic relationship classification: relationship type = GCN (concat ([knowledge point i vector, knowledge point j vector, time interval Δt])).

[0095] Step S203: determining the knowledge point importance channel of the knowledge point entity according to the global context.

[0096] Step S204: determining the behavior validity channel of the teaching behavior entity according to the behavior duration.

[0097] Step S205: The knowledge point importance channel and the behavior effectiveness channel are fused through a dual-channel graph attention network to obtain a joint influence weight.

[0098] Step S206: determining the teaching behavior association between the knowledge point entity and the teaching behavior entity according to the joint influence weight.

[0099] It is worth noting that the teaching behavior association modeling is performed here, and the influence weights of different teaching actions on knowledge mastery are calculated through the dual-channel graph attention network (Dual-GAT). Among them, the calculation formula of the first channel (knowledge point importance channel) is: α_i^K=softmax(W_K^T[h_i||h_{global}]), (h_i is the embedding of knowledge point i, h_{global} is the global context). The calculation formula of the second channel (behavior effectiveness channel) is: β_j^A=σ(W_A^T[a_j||Δt_j]), (a_j is the behavior feature, Δt_j is the behavior duration). The joint influence weight is calculated based on the above channel 1 and channel 2, and the formula is as follows:

[0100] w_{ij}=α_i^K×β_j^A×exp(-|t_i-t_j| / τ),

[0101] Where τ is the time decay coefficient, t_i is the start time of the behavior, and t_j is the end time of the behavior. Based on the joint influence weight obtained by the above calculation, the teaching behavior association between the knowledge point entity and the teaching behavior entity can be obtained.

[0102] Step S207: weightedly associate the knowledge point entity, the teaching behavior entity and the logical relationship according to the joint influence weight to obtain a classroom teaching knowledge graph.

[0103] It can be understood that the teaching behavior entities (for example, asking questions, demonstrating experiments, etc.) and the knowledge point entities are weighted and bound together based on logical relationships and joint influence weights to form a weighted teaching path map, that is, a classroom teaching knowledge map.

[0104] In a feasible implementation, step S207 may include steps S2071 to S2073:

[0105] Step S2071: construct a node according to the knowledge point entity and the teaching behavior entity.

[0106] Step S2072: establishing edges based on the logical relationship, the teaching behavior association, and the spatiotemporal proximity relationship between the nodes.

[0107] Step S2073, dynamically joint temporal modeling is performed based on the nodes and the edges through a temporal graph convolutional network and a long short-term memory network to generate a classroom teaching knowledge graph.

[0108] It's worth noting that to generate a weighted teaching path knowledge graph, we can first define the graph structure as follows: Node = {Knowledge Point Entity} ∪ {Teaching Behavior Entity}, Edge = Logical Relationship + Teaching Behavior Association + Spatiotemporal Proximity Relationship. We then use a Temporal Graph Convolutional Network (T-GCN) and a Long Short-Term Memory Network (LSTM) to jointly perform temporal modeling to predict classroom behavior trends, analyze changes in student engagement, and provide data support for dynamic classroom adjustments. The specific code is as follows:

[0109]

[0110] It should be noted that this paper also proposes an incremental graph optimization and real-time update mechanism. When a new knowledge point or behavior is detected, the following operations are triggered: ① Conflict detection: Calculate the cosine similarity between the new node and the existing nodes, and merge them if it exceeds the threshold; ② Path reinforcement: For repeated teaching logic chains, increase the edge weight: w_{ij} = w_{ij} + η*sigmoid (number of occurrences / time decay factor); ③ Redundancy pruning: Remove isolated nodes that have not been activated for 30 consecutive minutes.

[0111] In this embodiment, the knowledge point entity and the teaching behavior entity are identified based on the multimodal analysis results, and the logical relationship between the multiple knowledge point entities is extracted. The knowledge point importance channel of the knowledge point entity is determined based on the global context, and the behavior validity channel of the teaching behavior entity is determined based on the behavior duration. The knowledge point importance channel and the behavior validity channel are fused through a dual-channel graph attention network to obtain a joint influence weight. According to the joint influence weight, the knowledge point entity, the teaching behavior entity and the logical relationship are weightedly associated to obtain a classroom teaching knowledge graph, so that the relationship between knowledge points and teaching behaviors can be deeply analyzed, and a classroom teaching knowledge graph with higher correlation can be constructed.

[0112] Based on the first embodiment of the present application, in the third embodiment of the present application, the same or similar contents as those in the first embodiment can be referred to the above introduction and will not be described in detail later. Figure 3 , step S30, the classroom teaching ability evaluation method further includes steps S301 to S305:

[0113] Step S301: determine the similarity distance based on the shortest graph distance between the nodes in the classroom teaching knowledge graph and the preset expert knowledge graph.

[0114] It is understandable that graph preprocessing and node alignment are performed here, and semantic node matching is performed based on cross-graph alignment of knowledge representation. The knowledge embedding model TransE is used to map the nodes in the classroom teaching knowledge graph and the preset expert knowledge graph into a unified vector space:

[0115]

[0116] Among them, the classroom teaching knowledge graph is represented by G t =(V t ,E t ), the expert knowledge graph is represented as G e =(V e ,E e ), V represents a node, and E represents an edge.

[0117] It can be understood that the node similarity matrix is then calculated based on the TF-IDF algorithm:

[0118]

[0119] Then use the Hungarian Algorithm to perform optimal matching based on the similarity matrix:

[0120]

[0121] It should be noted that the spatiotemporal feature calibration (ve , v t ), calculate the space-time offset:

[0122]

[0123] Generate calibration weights based on spatiotemporal offsets:

[0124] w align =exp(-λ t Δt-λ s Δs).

[0125] It is understandable that after the above-mentioned graph preprocessing and node alignment are completed, the graph structure difference quantification is performed here. The similarity distance (Gromov-Wasserstein distance) is calculated using the improved Gromov-Wasserstein distance formula. The Gromov-Wasserstein distance is a distance function used to measure the similarity between two probability metric spaces. The calculation formula is as follows:

[0126]

[0127] Among them, (G t ,G e ) represents the classroom teaching knowledge graph and the expert graph, and T transmission matrix represents the correspondence between the teacher graph nodes and the expert graph nodes; T represents the feasible region of the transfer matrix; (d t ,d e ) represents the node distance of the teacher / expert graph (based on the shortest path in the graph), Represents a node in the teacher graph and distance; Nodes in the expert graph and distance; L(a,b)=|ab| 2 is the loss function.

[0128] Step S302: extracting paths with a frequency greater than a preset frequency in the classroom teaching knowledge graph according to a random walk algorithm to obtain high-frequency paths.

[0129] It is understandable that the random walk algorithm is used to extract high-frequency paths in the classroom teaching knowledge graph whose frequency is greater than the preset frequency:

[0130] P t ={(v1→v2→...→v n )}.

[0131] Step S303: determining the path difference between the high-frequency path and the key path library in the preset expert knowledge graph.

[0132] It should be understood that by comparing the high-frequency paths with the gold critical path library Pe of the expert graph, the path difference metric is obtained:

[0133]

[0134] Step S304: determining multi-dimensional ability evaluation indicators based on the classroom teaching knowledge graph and the preset expert knowledge graph.

[0135] In a feasible implementation, the multi-dimensional ability assessment indicator includes at least one of knowledge coverage, logical coherence, behavioral effectiveness, temporal and spatial rationality, and teacher-student interaction quality; step S304 may include steps S3041 to S3045:

[0136] Step S3041, determining the knowledge coverage according to the number of nodes in the classroom teaching knowledge graph and the number of nodes in the preset expert knowledge graph.

[0137] It should be noted that in this implementation, multi-dimensional ability evaluation indicators such as knowledge coverage, logical coherence, behavioral effectiveness, temporal and spatial rationality, and teacher-student interaction quality are calculated to provide a more comprehensive and accurate evaluation of teaching quality.

[0138] It is understandable that the knowledge coverage calculation formula is as follows:

[0139]

[0140] Among them, V t A node set representing the classroom teaching knowledge graph, Represents the core knowledge point set of the expert graph.

[0141] Step S3042, determining the logical coherence according to the number of cross-level jumps in the classroom teaching knowledge graph, the total number of knowledge points in the classroom teaching knowledge graph and the jump frequency benchmark value in the preset expert knowledge graph.

[0142] It is understandable that the formula for calculating logical coherence is as follows:

[0143]

[0144] Among them, N jump Indicates the number of cross-level jumps in the teacher graph, N total represents the total number of teacher graph knowledge points, Indicates the expert graph jump frequency benchmark value (usually ≈0.05).

[0145] Step S3043, determining the effectiveness of the behavior based on the probability distribution associated with the teaching behavior in the classroom teaching knowledge graph and the preset probability distribution in the preset expert knowledge graph.

[0146] It should be understood that the behavioral effectiveness calculation formula is as follows:

[0147]

[0148] Among them, D KL Used to measure the difference between two probability distributions; KL represents the divergence value; P t (x) represents the probability distribution of the association strength between behavior and knowledge points in the classroom teaching knowledge graph, P e (x) represents the probability distribution of the association strength between behavior and knowledge points in the expert graph; x represents the behavior-knowledge point association event; X The set of all possible behavior-knowledge point association events.

[0149] Step S3044, determining the spatiotemporal rationality based on the spatiotemporal distribution of nodes in the classroom teaching knowledge graph and the spatiotemporal distribution of nodes in the preset expert knowledge graph.

[0150] It should be understood that the calculation formula for spatiotemporal rationality is as follows:

[0151]

[0152] Among them, the time-space distribution gap between EDM teachers and experts; γ represents the transmission matrix; Γ(S t ,S e ) represents the set of all feasible transmission matrices; (S t ,S e ) represent the spatiotemporal distribution of teacher and expert graph nodes respectively; is the spatiotemporal transfer cost function, which means that the teacher node Adjust to expert node The cost of space-time location.

[0153] Step S3045, determining the teacher-student interaction quality based on the total number of interactive behaviors initiated by the teacher in the classroom teaching knowledge graph, the student participation rate of each interactive behavior, the interaction initiation time, and the preset importance in the preset expert knowledge graph.

[0154] It should be understood that the formula for calculating the quality of teacher-student interaction is as follows:

[0155]

[0156] Among them, N act represents the total number of interactive behaviors initiated by the teacher (questions, group discussions, etc.), R krepresents the student participation rate of the kth interaction (calculated by voice / video analysis), T k Indicates the duration of interaction initiation (unit: minutes, 0 at the beginning of the course), S k Indicates the importance of interactively associated knowledge points (derived from expert graph weights).

[0157] Step S305 : generating a comparison result according to the similarity distance, the path difference, and the multi-dimensional capability evaluation index.

[0158] It's important to note that a teaching ability radar chart can be constructed based on the five characteristic dimensions included in the multi-dimensional ability assessment indicators above. This chart identifies dimensions significantly below the expert benchmark (deviation >15%) and uses an isolation forest to detect anomalous teaching paths. Personalized teaching recommendations are then generated based on the assessment results. A hybrid architecture combining a rule engine and large-scale model fine-tuning can be employed to generate natural language recommendations based on the comparison results.

[0159] It is understandable that the rule engine (precise control) code example is as follows:

[0160]

[0161] Example output from the rule engine: "It was detected that the core knowledge point 'Proof of the Pythagorean Theorem' was not explained. Please refer to the 'Demonstration Derivation on Page 32 of the Textbook'."

[0162] It is understandable that the code example of large model generation (semantic expansion) is as follows:

[0163] "Based on LoRA fine-tuned LLM (such as LLaMA-3-Edu), input structured prompts:

[0164]

[0165]

[0166] Example output generated by the large model: "When explaining 'Metaphase of mitosis,' you can insert the question: 'How does the chromosome arrangement differ from prophase?' and then initiate a 2-minute group discussion. Historical data shows that this strategy can increase student comprehension by 23% (see case video clip_0045).

[0167] It should be noted that the above examples are only used to understand this application and do not constitute a limitation on the classroom teaching ability assessment method of this application. More simple transformations based on this technical concept are all within the scope of protection of this application.

[0168] This application also provides a classroom teaching ability assessment device, please refer to Figure 4, the classroom teaching ability evaluation device includes:

[0169] The multimodal analysis module 10 is used to analyze the classroom teaching video of the teacher to be evaluated based on the multimodal large language model to obtain a multimodal analysis result;

[0170] A knowledge graph construction module 20, configured to construct a classroom teaching knowledge graph based on the multimodal analysis results;

[0171] A knowledge graph comparison module 30 is used to compare the classroom teaching knowledge graph with a preset expert knowledge graph to obtain a comparison result;

[0172] The teaching ability evaluation module 40 is used to evaluate the classroom teaching ability of the teacher to be evaluated based on the comparison result to obtain an evaluation result.

[0173] The classroom teaching ability assessment device provided in this application utilizes the classroom teaching ability assessment method described in the above-mentioned embodiments to resolve the technical problem. Compared to the prior art, the beneficial effects of the classroom teaching ability assessment device provided in this application are the same as those of the classroom teaching ability assessment method described in the above-mentioned embodiments. The other technical features of the classroom teaching ability assessment device are the same as those disclosed in the above-mentioned embodiments, and are not further elaborated here.

[0174] The present application provides a classroom teaching ability assessment device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the classroom teaching ability assessment method in the above-mentioned embodiment one.

[0175] Reference below Figure 5 , which shows a schematic diagram of the structure of a classroom teaching ability assessment device suitable for implementing the embodiments of the present application. The classroom teaching ability assessment device in the embodiments of the present application may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), in-vehicle terminals (such as in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 5 The classroom teaching ability assessment device shown is merely an example and should not limit the functions and scope of use of the embodiments of the present application.

[0176] like Figure 5 As shown, the classroom teaching ability assessment device may include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes based on programs stored in a read-only memory 1002 or programs loaded from a storage device 1003 into a random access memory 1004. The random access memory 1004 also stores various programs and data required for the operation of the classroom teaching ability assessment device. The processing device 1001, the read-only memory 1002, and the random access memory 1004 are connected to each other via a bus 1005. An input / output interface 1006 is also connected to the bus. Typically, the following systems can be connected to the input / output interface 1006: an input device 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the classroom teaching ability assessment device to communicate wirelessly or wired with other devices to exchange data. Although the figure shows a classroom teaching ability assessment device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or provided instead.

[0177] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program comprising program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0178] The classroom teaching ability assessment device provided in this application, which employs the classroom teaching ability assessment method described in the above-mentioned embodiment, can solve the technical problems associated with classroom teaching ability assessment. Compared to the prior art, the beneficial effects of the classroom teaching ability assessment device provided in this application are the same as those of the classroom teaching ability assessment method described in the above-mentioned embodiment. The other technical features of the classroom teaching ability assessment device are the same as those disclosed in the above-mentioned embodiment, and are not further elaborated here.

[0179] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0180] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0181] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the classroom teaching ability evaluation method in the above-mentioned embodiment.

[0182] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0183] The above-mentioned computer-readable storage medium may be included in the classroom teaching ability evaluation device; or it may exist independently without being assembled into the classroom teaching ability evaluation device.

[0184] The above-mentioned computer-readable storage medium carries one or more programs. When the above-mentioned one or more programs are executed by the classroom teaching ability evaluation device, the classroom teaching ability evaluation device enables the following: to analyze the classroom teaching video of the teacher to be evaluated according to the multimodal large language model to obtain a multimodal analysis result; to construct a classroom teaching knowledge graph according to the multimodal analysis result; to compare the classroom teaching knowledge graph with the preset expert knowledge graph to obtain a comparison result; and to evaluate the classroom teaching ability of the teacher to be evaluated according to the comparison result to obtain an evaluation result.

[0185] Computer program code for performing the operations of the present application may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages such as Java, Smalltalk, C++, and conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0186] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, program segment or a part of code, and the module, program segment or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented by a dedicated hardware-based system that performs the specified function or operation, or can be implemented by a combination of dedicated hardware and computer instructions.

[0187] The modules described in the embodiments of the present application may be implemented in software or hardware, wherein the name of a module does not necessarily limit the unit itself.

[0188] The computer-readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described classroom teaching ability assessment method, thereby resolving the technical problem. Compared to the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the classroom teaching ability assessment method provided in the above-described embodiment, and are not further elaborated here.

[0189] The present application also provides a computer program product, including a computer program, which implements the steps of the above-mentioned classroom teaching ability assessment method when executed by a processor.

[0190] The computer program product provided in this application can solve technical problems. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the classroom teaching ability assessment method provided in the above embodiment, and will not be repeated here.

[0191] The above description is only part of the embodiments of the present application and does not limit the patent scope of the present application. All equivalent structural transformations made by using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect application in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A method for evaluating classroom teaching ability, characterized in that: The method includes: Analyze the classroom teaching videos of the teachers to be evaluated based on the multimodal large language model to obtain multimodal analysis results; Constructing a classroom teaching knowledge graph based on the multimodal analysis results; Comparing the classroom teaching knowledge graph with the preset expert knowledge graph to obtain a comparison result; The classroom teaching ability of the teacher to be evaluated is evaluated according to the comparison result to obtain an evaluation result.

2. The method according to claim 1, wherein The step of parsing the classroom teaching video of the teacher to be evaluated based on the multimodal large language model to obtain the multimodal parsing result includes: Conduct semantic analysis on the classroom speech data in the classroom teaching videos of the teachers to be evaluated to obtain classroom knowledge points; Detecting the teacher, students, and blackboard area to be evaluated based on the classroom teaching video to obtain teaching behaviors, wherein the teaching behaviors include at least one of classroom blackboard writing, teacher behavior, teacher expression, student behavior, and student expression; Multimodal feature alignment is performed on the classroom knowledge points and the teaching behaviors according to a multimodal large language model to obtain a multimodal analysis result.

3. The method according to claim 2, wherein The step of constructing a classroom teaching knowledge graph based on the multimodal analysis results includes: Identifying knowledge point entities and teaching behavior entities according to the multimodal analysis results; Extracting logical relationships between multiple knowledge point entities; Determining a knowledge point importance channel of the knowledge point entity according to a global context; Determine the behavior validity channel of the teaching behavior entity according to the behavior duration; The knowledge point importance channel and the behavior effectiveness channel are fused through a dual-channel graph attention network to obtain a joint influence weight; Determining the teaching behavior association between the knowledge point entity and the teaching behavior entity according to the joint influence weight; The knowledge point entity, the teaching behavior entity, the teaching behavior association and the logical relationship are weightedly associated according to the joint influence weight to obtain a classroom teaching knowledge graph.

4. The method according to claim 3, wherein The step of weightedly associating the knowledge point entity, the teaching behavior entity, the teaching behavior association, and the logical relationship according to the joint influence weight to obtain a classroom teaching knowledge graph includes: Constructing a node according to the knowledge point entity and the teaching behavior entity; Establishing edges according to the logical relationship, the teaching behavior association and the spatiotemporal proximity relationship between the nodes; Dynamic joint temporal modeling is performed based on the nodes and the edges through a temporal graph convolutional network and a long short-term memory network to generate a classroom teaching knowledge graph.

5. The method according to claim 4, wherein The step of comparing the classroom teaching knowledge graph with the preset expert knowledge graph to obtain a comparison result includes: Determining the similarity distance based on the shortest graph distance between nodes in the classroom teaching knowledge graph and the preset expert knowledge graph; Extracting paths with a frequency greater than a preset frequency in the classroom teaching knowledge graph according to a random walk algorithm to obtain high-frequency paths; Determining the path difference between the high-frequency path and the key path library in the preset expert knowledge graph; Determine multi-dimensional ability evaluation indicators based on the classroom teaching knowledge graph and the preset expert knowledge graph; A comparison result is generated according to the similarity distance, the path difference and the multi-dimensional capability evaluation index.

6. The method according to claim 5, wherein The multi-dimensional ability assessment indicators include at least one of knowledge coverage, logical coherence, behavioral effectiveness, temporal and spatial rationality, and teacher-student interaction quality; The step of determining the multi-dimensional ability evaluation indicators based on the classroom teaching knowledge graph and the preset expert knowledge graph includes: Determining knowledge coverage based on the number of nodes in the classroom teaching knowledge graph and the number of nodes in the preset expert knowledge graph; Determine the logical coherence according to the number of cross-level jumps in the classroom teaching knowledge graph, the total number of knowledge points in the classroom teaching knowledge graph, and the jump frequency benchmark value in the preset expert knowledge graph; Determining the effectiveness of the behavior based on the probability distribution of the teaching behavior association in the classroom teaching knowledge graph and the preset probability distribution in the preset expert knowledge graph; Determine spatiotemporal rationality based on the spatiotemporal distribution of nodes in the classroom teaching knowledge graph and the spatiotemporal distribution of nodes in the preset expert knowledge graph; The quality of teacher-student interaction is determined based on the total number of interactive behaviors initiated by the teacher in the classroom teaching knowledge graph, the student participation rate of each interactive behavior, the duration of interaction initiation, and the preset importance in the preset expert knowledge graph.

7. A classroom teaching ability assessment device, characterized in that: The classroom teaching ability evaluation device comprises: The multimodal parsing module is used to parse the classroom teaching videos of the teachers to be evaluated based on the multimodal large language model to obtain multimodal parsing results; A knowledge graph construction module, used to construct a classroom teaching knowledge graph based on the multimodal analysis results; A knowledge graph comparison module is used to compare the classroom teaching knowledge graph with the preset expert knowledge graph to obtain a comparison result; The teaching ability evaluation module is used to evaluate the classroom teaching ability of the teacher to be evaluated based on the comparison result to obtain an evaluation result.

8. A classroom teaching ability assessment device, characterized in that: The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the classroom teaching ability evaluation method according to any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the classroom teaching ability evaluation method according to any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The computer program product includes a computer program, and when the computer program is executed by a processor, the steps of the classroom teaching ability evaluation method according to any one of claims 1 to 6 are implemented.

Images