University VR teacher training simulation teaching data quantitative evaluation method and system

Abstract

The invention relates to the technical field of intelligent teaching, and discloses a college VR teacher training simulation teaching data quantitative evaluation method and system, and the method comprises the steps: carrying out the multi-modal feature fusion of the teaching behavior data of a teacher, and obtaining the teaching behavior features; performing semantic network reconstruction on the teaching behavior features to obtain a teaching behavior feature map; according to node attributes of the teaching behavior characteristic atlas, performing characteristic dimension mapping on the teaching behavior characteristic atlas to obtain teaching ability quantitative indexes; analyzing a dynamic evolution rule of the teaching ability quantitative index to obtain a change trend of the teaching ability quantitative index; performing stage trajectory fitting on the teaching ability quantitative index to obtain a teaching ability development trajectory; integrating the teaching ability quantitative index and the teaching ability development track to generate a comprehensive quantitative evaluation index; the college VR teacher training simulation teaching data quantitative evaluation method can improve the efficiency of college VR teacher training simulation teaching data quantitative evaluation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent teaching, in particular to a college VR teacher training simulation teaching data quantitative evaluation method and system. BACKGROUND

[0002] In the field of college VR teacher training simulation teaching data quantitative evaluation, the existing technology has significant defects in processing and feature extraction of teaching behavior data. The existing method is difficult to effectively fuse and semantically align the multi-modal teaching behavior data generated in the VR scene, often resulting in incomplete feature extraction or redundant features not being removed, which leads to the teaching behavior features obtained being unable to comprehensively and accurately reflect the actual teaching performance of teachers. At the same time, the existing technology lacks the ability of semantic association analysis and network reconstruction of teaching behavior features, and cannot construct an effective feature map, so that the calculation of subsequent teaching ability quantitative indicators lacks reliable data support, and the accuracy and objectivity of the evaluation results are difficult to guarantee.

[0003] In addition, the existing technology has obvious deficiencies in teaching ability dynamic evaluation and comprehensive index generation. Traditional evaluation methods focus on the calculation of static quantitative indicators, ignoring the dynamic evolution law of teaching ability with the training process, and cannot capture the change trend and stage characteristics of the indicators, making it difficult to generate the development trajectory of the teaching ability of teachers, resulting in evaluation results only reflecting the instantaneous teaching level, and unable to provide a basis for long-term training effect monitoring. At the same time, when integrating quantitative indicators and development trajectories to generate comprehensive evaluation results, the existing technology does not scientifically handle the dimension difference and weight configuration problems, which easily leads to distortion of the comprehensive index, and cannot accurately match the teaching ability evaluation standard, making it difficult to meet the demand of scientificity and practicality of the evaluation results of college VR teacher training. Therefore, how to improve the accuracy, dynamics and comprehensiveness of college VR teacher training simulation teaching data quantitative evaluation has become a problem to be solved. SUMMARY

[0004] The present application provides a college VR teacher training simulation teaching data quantitative evaluation method and system to solve the problems raised in the background art.

[0005] To achieve the above-mentioned purpose, the present application provides a college VR teacher training simulation teaching data quantitative evaluation method, which comprises:

[0006] S1, multi-modal feature fusion is performed on the teaching behavior data of teachers to obtain teaching behavior features of the teaching behavior data;

[0007] S2, the teaching behavior features are subjected to semantic network reconstruction to obtain a teaching behavior feature map of the teaching behavior features;

[0008] S3, mapping the teaching behavior feature graph according to the node attribute of the teaching behavior feature graph to obtain a teaching ability quantitative index of the teacher;

[0009] S4, analyzing the dynamic evolution law of the teaching ability quantitative index to obtain a change trend of the teaching ability quantitative index;

[0010] S5, fitting the teaching ability quantitative index according to the key features of the change trend to obtain a teaching ability development trajectory of the teacher;

[0011] S6, integrating the teaching ability quantitative index and the teaching ability development trajectory to generate a comprehensive quantitative evaluation index of the teacher.

[0012] In a preferred embodiment, the multi-modal feature fusion of the teaching behavior data of the teacher is performed to obtain the teaching behavior features of the teaching behavior data, including:

[0013] Obtaining the teaching behavior data of the teacher;

[0014] Extracting the speech features of the original speech waveform in the teaching behavior data;

[0015] Tracking the skeleton key point data of the teaching behavior data to obtain the posture features of the teacher;

[0016] Extracting the interaction frequency, response time and interaction depth of the teaching behavior data to obtain the interaction features of the teacher;

[0017] Aligning the speech features, the posture features and the interaction features to obtain the preliminary teaching behavior features of the teacher;

[0018] Eliminating the redundant features of the preliminary teaching behavior features to obtain the teaching behavior features of the teaching behavior data.

[0019] In a preferred embodiment, the semantic network reconstruction of the teaching behavior features is performed to obtain the teaching behavior feature graph of the teaching behavior features, including:

[0020] Performing semantic correlation analysis on the teaching behavior features to generate a feature similarity matrix of the teaching behavior features;

[0021] According to the correlation strength in the feature similarity matrix, when the correlation strength reaches a preset threshold, a connection relationship is established between the related features of the teaching behavior features;

[0022] Taking the related features as nodes and the connection relationship as connection edges, an initial feature graph of the teaching behavior features is constructed.

[0023] Semantic relation enhancement is performed on the initial feature map to obtain an optimized feature map.

[0024] Verify the connectivity of the optimized feature map to confirm the teaching behavior feature map of the teaching behavior features.

[0025] In a preferred embodiment, the step of performing semantic association analysis on the teaching behavior features to generate a feature similarity matrix of the teaching behavior features includes:

[0026] The teaching behavior features are numerically processed while retaining their semantic information to obtain feature vectors.

[0027] Based on the distribution characteristics of the feature vectors, construct the semantic space framework of the feature vectors;

[0028] Within the semantic space framework, the relative positions and distribution patterns of the feature vectors are analyzed to obtain the similarity relationship data of the feature vectors.

[0029] The similarity relationship data is organized into a matrix structure to obtain the feature similarity matrix of the teaching behavior features.

[0030] In a preferred embodiment, the step of mapping the teaching behavior feature map to feature dimensions based on the node attributes of the teaching behavior feature map to obtain the teacher's quantitative teaching ability index includes:

[0031] Collect the node attribute parameters of the teaching behavior feature map;

[0032] Based on the semantic relevance of the node attribute parameters, the node attribute parameters are semantically associated and classified to obtain the dimensional grouping data of the node attribute parameters;

[0033] The dimensional grouped data is mapped into a vector space to obtain the dimensional feature vectors of the dimensional grouped data.

[0034] Based on the preset feature weight configuration, the quantization score of the dimensional feature vector is calculated, wherein the calculation formula for the quantization score is as follows: ;

[0035] In the formula, This represents the quantized score of the feature vector of the stated dimension. Indicates the first The weight coefficients of the feature vectors in each dimension. The dimensional feature vector represents the first... Each component value represents a Sigmoid activation function, represents a historical mean value of a component in the dimension feature vector, represents a preset information entropy adjustment coefficient, represents an information entropy of the dimension feature vector, represents a dimension number of the dimension feature vector, represents a summation operation, represents a square root operation;

[0036] mapping the quantitative score to a preset teaching ability evaluation standard to obtain a teaching ability quantitative indicator of the teacher.

[0037] In a preferred embodiment, the analysis of the dynamic evolution law of the teaching ability quantitative indicator to obtain the change trend of the teaching ability quantitative indicator comprises: serializing the teaching ability quantitative indicator to obtain a time sequence indicator sequence of the teaching ability quantitative indicator;

[0038] performing dynamic mode analysis on the time sequence indicator sequence to obtain a trend mode of the time sequence indicator sequence;

[0039] According to the trend mode, the trend of the teaching ability quantitative indicator is synthesized to obtain the change trend of the teaching ability quantitative indicator.

[0040] In a preferred embodiment, the according to the key features of the change trend, the stage trajectory fitting of the teaching ability quantitative indicator to obtain the teaching ability development trajectory of the teacher comprises:

[0041] extracting the key features of the change trend;

[0042] According to the time distribution and feature intensity of the key features, the state serialization of the teaching ability quantitative indicator is performed to obtain an evolution state sequence of the teaching ability quantitative indicator;

[0043] According to the logical association between states in the evolution state sequence, a connection relationship between the states is established;

[0044] Taking the states in the evolution state sequence as trajectory nodes and the connection relationship as connection edges, a basic trajectory framework of the teaching ability quantitative indicator is constructed;

[0045] Performing segmented trajectory interpolation on the basic trajectory framework to obtain a segmented trajectory segment of the basic trajectory framework;

[0046] Performing transition zone smoothing on the segmented trajectory segment to obtain the teaching ability development trajectory of the teacher.​​

[0047] In a preferred embodiment, the transition zone smoothing of the segmented trajectory segment obtains the teaching ability development trajectory of the teacher, comprising:

[0048] According to the connection point characteristics of the segmented trajectory segment, the transition zone region of the segmented trajectory segment is identified;

[0049] The curvature continuity optimization is performed on the trajectory points in the transition zone region to obtain the smoothed trajectory of the segmented trajectory segment;

[0050] The smoothness of the smoothed trajectory is verified to confirm the teaching ability development trajectory of the teacher.

[0051] In a preferred embodiment, the integration of the teaching ability quantitative index and the teaching ability development trajectory generates the comprehensive quantitative evaluation index of the teacher, comprising:

[0052] According to the target requirements of the teaching ability evaluation standard, the weight coefficients of the teaching ability quantitative index and the teaching ability development trajectory are configured;

[0053] Based on the weight coefficients, the teaching ability quantitative index and the teaching ability development trajectory are index fused to obtain the preliminary comprehensive index of the teacher;

[0054] The dimensional influence of the preliminary comprehensive index is eliminated to obtain the comprehensive quantitative evaluation index of the teacher.

[0055] In order to solve the above problems, the application also provides a college VR faculty training simulation teaching data quantitative evaluation system, the system comprises:

[0056] A multi-modal feature fusion module is configured to perform multi-modal feature fusion on the teaching behavior data of a teacher to obtain teaching behavior characteristics of the teaching behavior data;

[0057] A semantic network reconstruction module is configured to perform semantic network reconstruction on the teaching behavior characteristics to obtain a teaching behavior characteristic map of the teaching behavior characteristics;

[0058] A feature dimension mapping module is configured to perform feature dimension mapping on the teaching behavior characteristic map according to the node attributes of the teaching behavior characteristic map to obtain a teaching ability quantitative index of the teacher;

[0059] A dynamic evolution analysis module is configured to analyze the dynamic evolution law of the teaching ability quantitative index to obtain the change trend of the teaching ability quantitative index;

[0060] a stage trajectory fitting module, configured to perform stage trajectory fitting on the teaching ability quantitative indicators according to the key features of the change trend, to obtain a teaching ability development trajectory of the teacher;

[0061] a comprehensive evaluation index generation module, configured to integrate the teaching ability quantitative indicators and the teaching ability development trajectory, to generate a comprehensive quantitative evaluation index of the teacher.

[0062] Compared with the prior art, the present application has the following beneficial effects:

[0063] 1. The present application can comprehensively extract effective information in the teacher teaching behavior data and accurately generate a teaching behavior feature map through multi-modal feature fusion and semantic network reconstruction technology. It can integrate multi-dimensional features such as voice, posture and interaction and eliminate redundant information to ensure the integrity and accuracy of the teaching behavior features; at the same time, it relies on the feature similarity matrix to construct a closely related feature map, providing reliable data support for the teaching ability quantitative indicators, significantly improving the matching degree of the quantitative indicators and the actual teaching ability of the teacher, and enhancing the objectivity and accuracy of the evaluation results.

[0064] 2. The present application can clearly present the change trend and development trajectory of the teaching ability quantitative indicators through dynamic evolution analysis and stage trajectory fitting, realize dynamic monitoring and long-term tracking of the teaching ability of the teacher; and generate accurate comprehensive quantitative evaluation indicators through scientific weight configuration and dimensionless processing of the integrated indicators and trajectories. This not only can intuitively reflect the ability improvement process of the teacher in VR training, but also can provide a clear direction for training program optimization, effectively improve the scientificity and practicality of the evaluation of VR teacher training in colleges and universities, and help to efficiently cultivate high-quality teachers. BRIEF DESCRIPTION OF DRAWINGS

[0065] Figure 1 a flowchart of a college VR teacher training simulation teaching data quantitative evaluation method provided by an embodiment of the present application;

[0066] Figure 2 a functional module diagram of a college VR teacher training simulation teaching data quantitative evaluation system provided by an embodiment of the present application;

[0067] The implementation, functional characteristics and advantages of the present application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0068] It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0069] The embodiment of the application provides a college VR teacher training simulation teaching data quantitative evaluation method. The execution subject of the college VR teacher training simulation teaching data quantitative evaluation method includes but is not limited to at least one of the electronic devices capable of being configured to execute the method provided by the embodiment of the application, such as a server and a terminal. In other words, the college VR teacher training simulation teaching data quantitative evaluation method can be executed by software or hardware installed in a terminal device or a server device. The server includes but is not limited to a single server, a server cluster, a cloud server or a cloud server cluster, etc. The server can be a stand-alone server, or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content distribution networks (CDN), and big data and artificial intelligence platforms, etc. Basic cloud computing services.

[0070] Reference Figure 1 As shown in the figure, the embodiment of the application provides a college VR teacher training simulation teaching data quantitative evaluation method. In this embodiment, the college VR teacher training simulation teaching data quantitative evaluation method includes:

[0071] S1, multi-modal feature fusion is performed on the teaching behavior data of a teacher to obtain teaching behavior features of the teaching behavior data;

[0072] In the embodiment of the application, the multi-modal feature fusion is performed on the teaching behavior data of the teacher to obtain the teaching behavior features of the teaching behavior data, including:

[0073] Obtaining the teaching behavior data of the teacher;

[0074] Extracting speech features of original speech waveforms in the teaching behavior data;

[0075] Tracking the skeleton key point data of the teaching behavior data to obtain the posture features of the teacher;

[0076] Extracting features of the interaction frequency, response time and interaction depth of the teaching behavior data to obtain the interaction features of the teacher;

[0077] Performing semantic alignment on the speech features, the posture features and the interaction features to obtain preliminary teaching behavior features of the teacher;

[0078] Eliminating redundant features of the preliminary teaching behavior features to obtain the teaching behavior features of the teaching behavior data.

[0079] When obtaining the teaching behavior data of the teacher, through the built-in data acquisition module of the VR teacher training simulation teaching system of the university, all behavior-related data of the teacher in the VR teaching scene are captured in real time, including the voice data, body action data generated by the teacher in the teaching process, and the interaction data with the virtual teaching object or the virtual teaching scene. These data are stored in the system database in a structured format, ensuring that the teaching behavior data required for subsequent processing is complete and callable.

[0080] When extracting the voice features of the original voice waveform in the teaching behavior data, the stored original voice waveform data is preprocessed to remove background noise in the voice, and then the voice signal analysis tool is used to process the voice waveform frame by frame to extract the frequency, amplitude, speech rate, and voice pause interval of each frame of voice. These extracted features are integrated to form the voice features of the teacher and stored in the feature database.

[0081] When tracking the skeleton key point data of the teaching behavior data to obtain the posture features of the teacher, the motion capture device equipped with the VR system is used to locate the key skeleton nodes of the teacher's body in real time, including the head, neck, shoulder, elbow, wrist, hip, knee, and ankle nodes. The three-dimensional coordinate changes of each key node in the VR scene are recorded, and the amplitude of the teacher's body movement, the continuity of the movement, and the expression state of the body language are analyzed according to these coordinate changes to integrate the analysis results to form the posture features of the teacher.

[0082] When extracting the interaction frequency, response time, and interaction depth of the teaching behavior data to obtain the interaction features of the teacher, the number of interactions between the teacher and the virtual teaching object or the virtual teaching scene per unit time is first counted to determine the interaction frequency. Then, the time interval from the issuance of the interaction demand by the virtual teaching scene or the virtual teaching object to the response of the teacher is recorded, and the average response time is calculated. At the same time, the content integrity of each interaction, the satisfaction degree of the interaction demand, and the guidance in the interaction process are analyzed to determine the interaction depth. The related data of the interaction frequency, response time, and interaction depth are integrated to form the interaction features of the teacher.

[0083] When performing semantic alignment on the voice features, posture features, and interaction features, a unified semantic reference system is first established to clearly define the semantic correspondence of the three features in the teaching behavior description, such as establishing semantic association between "question tone" in voice features and "gesture pointing action" in posture features, and "response interaction to the question" in interaction features. Then, according to the semantic reference system, the time dimension and description dimension of the three features are adjusted to make the three features consistent in time node and semantic expression, forming the preliminary teaching behavior features of the teacher.

[0084] When removing redundant features of the preliminary teaching behavior characteristics, the relevance between each feature item in the preliminary teaching behavior characteristics is analyzed first, and the feature items that repeatedly describe the same teaching behavior and the feature items that have no substantial impact on the teaching ability evaluation are identified. For example, the preliminary characteristics contain both "hand lifting amplitude" and "upper limb lifting amplitude", and the contents of the two are highly coincident. Therefore, the more representative "upper limb lifting amplitude" feature item is retained, and the "hand lifting amplitude" feature item is deleted. After screening, the teaching behavior characteristics of the teaching behavior data are obtained.

[0085] The teaching behavior data of the teacher is obtained, including the skeletal key point trajectory data collected in the VR teaching scene, the original speech waveform data, and the interactive behavior record data, including the number of interactive triggers, the response time, and the interactive content correlation score.

[0086] The speech features of the original speech waveform in the teaching behavior data are extracted, including speech speed, logical pause interval, volume fluctuation amplitude, and speech intelligibility score, which are calculated by the mel-frequency cepstral coefficient. These speech features directly reflect the language expression and classroom communication ability of the teacher. For example, when the speech speed is stable at 50 to 70 words per minute, and the logical pause interval and the matching degree of knowledge point division are not less than 80%, it indicates that the teaching language organization ability is good.

[0087] The skeletal key point data of the teaching behavior data are tracked, and the core nodes include the head, neck, shoulder, elbow, wrist, hip, knee, and ankle. The posture features of the teacher are obtained by calculating the three-dimensional coordinate change, the motion coherence coefficient, and the pointing accuracy. These posture features directly reflect the body expression and classroom demonstration ability of the teacher. For example, when the hand node fluctuation range is not more than 5 degrees and the pointing deviation angle is not more than 3 degrees during experimental operation demonstration, it indicates that the teaching operation demonstration standardization meets the standard.

[0088] The interaction frequency, response time, and interaction depth of the teaching behavior data are extracted to obtain the interaction features of the teacher. These interaction features directly reflect the classroom interaction and emergency feedback ability of the teacher. For example, when the interaction frequency is not less than 0.5 times per minute within 10 minutes, the average response time is not more than 2 seconds, and the interaction depth score is not less than 7 points, it indicates that the teaching interaction guidance ability is strong.

[0089] The speech features, posture features, and interaction features are semantically aligned, the collection nodes of the three types of features are synchronized based on the timestamp, a semantic association mapping table of speech, posture, and interaction is established, and the preliminary teaching behavior characteristics of the teacher are obtained.

[0090] A feature screening algorithm based on mutual information is used to remove redundant features in the preliminary teaching behavior characteristics, i.e., features that repeatedly reflect the same teaching ability, and the teaching behavior characteristics of the teaching behavior data are obtained.

[0091] Taking the VR college physics experiment-circuit connection teacher training in colleges and universities as an example, the teaching behavior data of teacher A is collected. In terms of skeletal data, the three-dimensional coordinate change of hand wrist and fingertip key points is not more than 4 degrees, the motion continuity coefficient is 0.92, the deviation angle of pointing to circuit elements is not more than 2 degrees, and the corresponding posture feature quantization value is 8.6 points, reflecting that the experimental operation demonstration ability is at a high level. In terms of voice data, the speech speed is 62 words per minute, the average logical pause interval is 1.1 seconds, which is completely matched with the circuit connection step division, the volume fluctuation amplitude is not more than 3dB, the speech intelligibility score is 9.1 points, and the corresponding voice feature quantization value is 8.8 points, reflecting that the language expression clarity and logical organization ability are good. In terms of interaction data, 7 interactions are initiated within 10 minutes, the interaction frequency is 0.7 times per minute, the average response time of responding to virtual student questions is 1.6 seconds, the interaction depth score is 8.3 points, 80% of the responses can guide students to correct operation deviation, and the corresponding interaction feature quantization value is 8.2 points, reflecting that the interaction guidance and emergency feedback ability is strong. After semantic alignment and redundancy elimination, the teaching behavior characteristics of teacher A are finally obtained, and three types of directly assessable teaching abilities, namely, core quantization experimental operation demonstration ability, language expression and communication ability, and classroom interaction and emergency feedback ability, are obtained.

[0092] The beneficial effects are that through the specific implementation of the above multi-step, the speech features, posture features and interaction features can be comprehensively and accurately extracted from the teaching behavior data of the teacher, and through semantic alignment and redundant feature elimination, complete, accurate and non-redundant teaching behavior characteristics are obtained, which provides a high-quality data basis for subsequent construction of a teaching behavior feature map and generation of teaching ability quantization indexes, and ensures the reliability and accuracy of the subsequent evaluation link.

[0093] S2, the teaching behavior characteristics are subjected to semantic network reconstruction to obtain a teaching behavior feature map of the teaching behavior characteristics;

[0094] In the embodiment of the application, the teaching behavior characteristics are subjected to semantic network reconstruction to obtain a teaching behavior feature map of the teaching behavior characteristics, which comprises:

[0095] The teaching behavior characteristics are subjected to semantic correlation analysis to generate a feature similarity matrix of the teaching behavior characteristics;

[0096] According to the correlation strength in the feature similarity matrix, when the correlation strength reaches a preset threshold, a connection relationship is established between the related characteristics of the teaching behavior characteristics;

[0097] The related characteristics are taken as nodes, and the connection relationship is taken as a connection edge, to construct an initial feature map of the teaching behavior characteristics;

[0098] The initial feature map is subjected to semantic relationship enhancement to obtain an optimized feature map of the initial feature map.

[0099] The connectivity of the optimized feature map is verified to confirm a teaching behavior feature map of the teaching behavior features.

[0100] The semantic correlation analysis of the teaching behavior features generates a feature similarity matrix of the teaching behavior features, including:

[0101] The teaching behavior features are subjected to numerical value processing while retaining the semantic information of the teaching behavior features to obtain a feature vector of the teaching behavior features.

[0102] A semantic space framework of the feature vector is constructed according to the distribution characteristics of the feature vector.

[0103] In the semantic space framework, the relative position and distribution pattern of the feature vector are analyzed to obtain similarity relationship data of the feature vector.

[0104] The similarity relationship data is subjected to matrix structure organization to obtain a feature similarity matrix of the teaching behavior features.

[0105] When the teaching behavior features are subjected to numerical value processing while retaining the semantic information to obtain a feature vector, a corresponding quantization rule is set for the attribute of each teaching behavior feature, for example, the speech speed is converted into a specific numerical value according to "words per minute", and the motion amplitude is converted into a specific numerical value according to "limb movement angle range", and a corresponding semantic description is labeled for each quantized numerical value to ensure that the semantic information of the feature is retained after numerical value processing. Each processed teaching behavior feature is stored in the form of a vector containing quantized numerical values and semantic labels to obtain a feature vector of the teaching behavior features.

[0106] When the semantic space framework of the feature vector is constructed according to the distribution characteristics of the feature vector, the distribution range and frequency of each dimension value in all feature vectors are first counted to determine the feature performance of each feature vector in the numerical value dimension. Then, according to the semantic labels of the feature vectors, the feature vectors with the same or similar semantic types are classified into the same semantic category, for example, the feature vectors containing "speech speed" and "speech pause interval" are classified into the "speech semantic category", and the feature vectors containing "motion amplitude" and "motion continuity" are classified into the "posture semantic category". Based on these semantic categories and the numerical value distribution of the feature vectors, a space structure containing semantic category dimensions and numerical value distribution dimensions is built to form a semantic space framework of the feature vectors.

[0107] In the semantic space framework, the relative position and distribution pattern of the feature vector are analyzed to obtain the similarity relationship data. Each feature vector is mapped to the corresponding position of the semantic space framework. By observing the attribution of different feature vectors in the semantic category dimension and the distance in the numerical distribution dimension, the correlation between the feature vectors is determined. For example, if two feature vectors belong to the same "voice semantic category" and the distance in the numerical distribution dimension is less than the preset distance range, it is determined that the similarity is high. If two feature vectors belong to different semantic categories and the numerical distribution distance is large, it is determined that the similarity is low. The similarity determination results between all feature vectors are arranged in the form of pair records to obtain the similarity relationship data of the feature vectors.

[0108] The similarity relationship data is organized in a matrix structure to obtain the feature similarity matrix. The rows and columns of the matrix correspond to the feature vectors of all teaching behavior features. Each element position in the matrix corresponds to a pair of feature vectors. The similarity result of each pair of feature vectors in the similarity relationship data is filled into the corresponding element position in the matrix. For example, feature vector A corresponds to the first row of the matrix, and feature vector B corresponds to the second column of the matrix. The similarity result of A and B is filled into the element in the first row and the second column of the matrix. All similarity relationship data is filled into the matrix in this way to form the feature similarity matrix of the teaching behavior features.

[0109] In the semantic association analysis of the teaching behavior features, the above-mentioned numerical processing of the teaching behavior features to obtain the feature vector, the construction of the semantic space framework, the analysis of the similarity of the feature vectors to obtain the similarity relationship data, and the organization of the matrix structure to obtain the feature similarity matrix are performed. Through this process, the semantic association between the teaching behavior features is completely mined, and the feature similarity matrix reflecting the correlation strength between the features is finally generated.

[0110] According to the correlation strength in the feature similarity matrix, when the correlation strength reaches the preset threshold, a connection relationship is established between the related features. First, the correlation strength value corresponding to each element in the feature similarity matrix is read. The value is compared with the preset correlation strength threshold. If the correlation strength value corresponding to an element is greater than or equal to the preset threshold, it is determined that the two teaching behavior features corresponding to the element are related features. Then, a connection record representing the correlation relationship is established between the two related features. The record contains the identification of the two related features and the corresponding correlation strength information. The establishment of the connection relationship is completed.

[0111] In constructing the initial feature graph of teaching behavior features with relevant features as nodes and connection relationships as connection edges, first, all teaching behavior features determined as relevant features are taken as independent nodes, each node is labeled with the corresponding feature name and core attribute, and then the previously established connection relationships are taken as connection edges, the two relevant feature nodes corresponding to each connection edge are connected, and the corresponding association strength information is labeled on the connection edge. According to the corresponding relationship between the nodes and the edges, the nodes and the connection edges are added one by one on the graph construction platform to form the initial feature graph that can intuitively display the association relationship between the teaching behavior features. In the semantic relationship enhancement of the initial feature graph to obtain the optimized feature graph, first, the semantic annotations and attributes of each node in the initial feature graph are analyzed to identify the potential indirect semantic association between the nodes. For example, node A and node B have a direct connection, node B and node C have a direct connection, and node A and node C belong to the same “interaction-related feature” in semantics. A supplementary connection edge representing the indirect association is added between node A and node C, and the semantic basis of the indirect association is labeled. Then, the completeness of the association strength annotations of the connection edges in the initial graph is checked, and the missing or inaccurate annotations are corrected. After the supplementary association and the correction of the annotations, the optimized feature graph of the initial feature graph is obtained.

[0112] In verifying the connectivity of the optimized feature graph to confirm the teaching behavior feature graph of teaching behavior features, first, it is checked whether all nodes in the optimized feature graph can form a connected whole through the connection edges, that is, from any one node, all other nodes in the graph can be reached by traversing along the connection edges. If it is found that there is an isolated node that cannot be connected, the association strength between the isolated node and other nodes is analyzed. If the association strength between the isolated node and a certain node is close to a preset threshold, the association relationship between the two nodes is reevaluated and a connection edge is attempted to be supplemented. If the isolated node indeed has no associated feature meeting the threshold, it is marked and recorded separately. If all nodes can form a connected structure or the isolated nodes have been reasonably processed, it is confirmed that the optimized feature graph is the final teaching behavior feature graph.

[0113] The beneficial effects are that through the above specific implementation process, the semantic network reconstruction of teaching behavior features can be systematically and accurately completed from feature vector generation, similarity matrix construction, connection relationship establishment, graph construction and optimization. At each step, the semantic association of teaching behavior features is fully mined, and finally a teaching behavior feature graph with good connectivity and clear association relationship is obtained. The graph can intuitively display the association strength between each teaching behavior feature, providing structured and high-quality graph data support for subsequent extraction of teaching capability quantitative indicators based on node attributes, and ensuring the accuracy and reliability of the subsequent evaluation link.

[0114] S3. Based on the node attributes of the teaching behavior feature map, perform feature dimension mapping on the teaching behavior feature map to obtain the quantitative indicators of the teacher's teaching ability.

[0115] In this embodiment of the invention, the step of mapping the teaching behavior feature map to feature dimensions based on the node attributes of the teaching behavior feature map to obtain the quantitative indicators of the teacher's teaching ability includes:

[0116] Collect the node attribute parameters of the teaching behavior feature map;

[0117] Based on the semantic relevance of the node attribute parameters, the node attribute parameters are semantically associated and classified to obtain the dimensional grouping data of the node attribute parameters;

[0118] The dimensional grouped data is mapped into a vector space to obtain the dimensional feature vectors of the dimensional grouped data.

[0119] Based on the preset feature weight configuration, the quantization score of the dimensional feature vector is calculated, wherein the calculation formula for the quantization score is as follows:

[0120] ;

[0121] In the formula, This represents the quantized score of the feature vector of the stated dimension. Indicates the first The weight coefficients of the feature vectors in each dimension. The dimensional feature vector represents the first... Each component value This represents the Sigmoid activation function. Represents the dimensional feature vector of the th dimension Historical mean of each component, This represents the preset information entropy adjustment coefficient. Represents the dimensional feature vector Information entropy This represents the number of dimensions in the dimensional feature vector. This represents the summation operation. This represents the square root operation;

[0122] The quantitative score is mapped to a preset teaching ability assessment standard to obtain the teacher's quantitative teaching ability index.

[0123] When collecting the node attribute parameters of the teaching behavior feature graph, the attribute information corresponding to each node in the teaching behavior feature graph needs to be identified one by one. These attribute information includes the type description of the teaching behavior feature represented by the node, the performance frequency of the feature in the teaching process, the specific numerical performance corresponding to the feature, and the association strength of the feature with other nodes, etc. These attribute information of each node is recorded and sorted one by one to form the node attribute parameter set of the teaching behavior feature graph, ensuring that the attribute parameters of each node are not missed and accurately correspond to the node itself.

[0124] When the node attribute parameters are semantically associated and classified according to the semantic correlation of the node attribute parameters to obtain the dimension grouping data of the node attribute parameters, the semantic description of each node attribute parameter is analyzed first to determine the corresponding teaching ability dimension. For example, the parameters containing semantic descriptions such as "speech speed" and "speech pause interval" in the node attribute parameters are all related to the teaching ability dimension of "teacher language expression ability", and these parameters are classified into the same group. The parameters containing semantic descriptions such as "action amplitude" and "action continuity" are all related to the teaching ability dimension of "teacher body expression ability", and are classified into another group. The parameters containing semantic descriptions such as "interaction frequency" and "response time" are all related to the teaching ability dimension of "teacher classroom interaction ability", and are classified into a third group. In this way, all node attribute parameters are divided into different teaching ability dimension groups according to the semantic association judgment method, and each group is the dimension grouping data of the node attribute parameters.

[0125] When the dimension grouping data is mapped to a vector space to obtain the dimension feature vector of the dimension grouping data, the corresponding vector dimension is set for each dimension grouping data first, and the number of dimensions is consistent with the number of node attribute parameters in the group. Then, the numerical performance of each node attribute parameter in each group is taken as a component value of the vector, and these component values are arranged in order according to the preset order to form an ordered numerical sequence. This numerical sequence is the dimension feature vector corresponding to the dimension grouping data, ensuring that each dimension grouping data can be accurately converted into a unique corresponding dimension feature vector.

[0126] When calculating the quantization score of a dimensional feature vector based on a preset feature weight configuration, the weights corresponding to each component of the dimensional feature vector in the preset feature weight configuration are first obtained. Then, the difference between each component value and its historical mean is calculated. Each difference is substituted into the Sigmoid activation function for processing to obtain the normalized result corresponding to each component. Next, each normalized result is multiplied by the weight of the corresponding component. All multiplication results are added together to obtain the sum of the numerators. Then, the sum of the squares of all component weights is calculated and the square root is taken to obtain the first part of the denominator. Then, the information entropy of the dimensional feature vector is calculated. The information entropy is multiplied by a preset information entropy adjustment coefficient and then 1 is added to obtain the second part of the denominator. The first part of the denominator is multiplied by the second part to obtain the final denominator. Finally, the sum of the numerators is divided by the final denominator to obtain the quantization score of the dimensional feature vector.

[0127] The weight coefficients of the dimensional eigenvectors are determined using the analytic hierarchy process (AHP), with the following steps: First, a weight judgment matrix is ​​established. A review panel composed of three university teaching experts, two VR technology experts, and two teacher training managers compares and scores the importance of three dimensions—language expression, physical demonstration, and interactive guidance—based on the core objectives of VR teaching scenarios (e.g., experimental courses emphasize operational demonstrations, while theoretical courses emphasize verbal expression). A 1-9 scale is used to obtain the judgment matrix. Second, a consistency check is performed. The consistency index (CI) and consistency ratio (CR) of the judgment matrix are calculated. If CR is less than 0.1, the judgment matrix meets the consistency requirements; otherwise, the scores are recalculated. Third, weight calculation is performed. The eigenvectors of the judgment matrix are calculated using eigenvalue decomposition and normalized to obtain the weight coefficients for each dimension.

[0128] For the VR training course "University Physics Experiment - Circuit Connection," the core objective is to improve experimental operation demonstration and interactive guidance capabilities. The review panel determined the weighting coefficients after analysis using the Analytic Hierarchy Process (AHP). The physical demonstration dimension corresponds to experimental operation demonstration ability, with a weighting coefficient of 0.40. This is based on the fact that standardized operation demonstration is a core teaching objective in experimental courses and directly impacts the development of students' practical skills, thus it has the highest weight. The interactive guidance dimension corresponds to classroom interaction and emergency feedback capabilities, with a weighting coefficient of 0.35. This is based on the fact that experimental classes require timely responses to students' operational questions, and interactive ability directly determines the effectiveness of teaching, thus it has the second highest weight. The verbal expression dimension corresponds to verbal expression and communication skills, with a weighting coefficient of 0.25. This is based on the fact that language is a tool to assist in operation demonstrations; it must ensure clear expression but not supersede the operation, thus it has the third highest weight.

[0129] The formula for calculating the quantitative score is as follows: This represents the quantized score of the feature vector of the stated dimension, which, after normalization, ranges from 0 to 1. Indicates the first The weight coefficients of the feature vectors in each dimension. The dimensional feature vector represents the first... Each component value This represents the Sigmoid activation function, used to normalize component values ​​to the interval between 0 and 1. Represents the dimensional feature vector of the first dimension. The historical average of each component was obtained based on statistics from over 1000 sets of similar VR teaching data. This represents the preset information entropy adjustment coefficient, with a value range of 0.5 to 1.0 and a default of 0.8. Represents the dimensional feature vector Information entropy reflects the uniformity of feature distribution. This indicates the number of dimensions in the dimensional feature vector.

[0130] Using Teacher A's case, Equals 3, Equals 0.8, physical demonstration dimension It is 0.40. It is 8.6. It is 7.5. reduce It equals 1.1. Equals 0.750, Interactive Guidance Dimension It is 0.35. It is 8.2. It is 6.8. reduce It equals 1.4. Equals 0.802, Language Expression Dimension It is 0.25. It is 8.8. It is 7.2. reduce It equals 1.6. The result is 0.833; the numerator is calculated as 0.40 × 0.750 + 0.35 × 0.802 + 0.25 × 0.833, which gives 0.300 + 0.2807 + 0.2083 = 0.789; the denominator is calculated by first calculating the sum of squares of the weights, which gives 0.345; the square root of the sum of squares of the weights is approximately 0.587; information entropy. The entropy value of the probability distribution of the dimensional feature vector was calculated, and the result was 1.2, reflecting good uniformity of the feature distribution; the adjustment term was 0.8 × 1.2 + 1 = 1.96; the final denominator value was 0.587 × 1.96 ≈ 1.151; the quantized score was... It equals 0.789 divided by 1.151, which is approximately 0.686. After normalization, it is converted to a percentage score of 68.6.

[0131] The quantitative score is mapped to the preset teaching ability evaluation standard to obtain the quantitative index of the teaching ability of the teacher. The core logic is "threshold value based on big data statistical law → score matching level → linear transformation index". First, the teaching ability level corresponding to the different score intervals in the preset teaching ability evaluation standard and the specific quantitative index are determined. For example, the quantitative score in the evaluation standard is 80-100, which corresponds to the "excellent" level, and the corresponding teaching ability quantitative index is 90-100; 60-79 corresponds to the "good" level, and the corresponding teaching ability quantitative index is 70-89; 40-59 corresponds to the "qualified" level, and the corresponding teaching ability quantitative index is 50-69; and 40 or less corresponds to the "unqualified" level, and the corresponding teaching ability quantitative index is 0-49. Then, the interval in which the calculated quantitative score is located is found, and the teaching ability quantitative index range corresponding to the interval is found. According to the specific position of the quantitative score in the interval, the final teaching ability quantitative index value is determined. For example, the quantitative score is 85, which is in the 80-100 interval, corresponding to the "excellent" level. If the quantitative score and the teaching ability quantitative index in the interval are linearly mapped by 1.10 times, then the quantitative score corresponding to the teaching ability quantitative index is 93.5, which is determined as the quantitative index of the teaching ability of the teacher.

[0132] The threshold value of the teaching ability evaluation standard is determined based on the statistical distribution law of 1000 or more VR teacher training sample data. The sample data conforms to the normal distribution, and the threshold value is the mean value plus or minus 1 standard deviation. The specific mapping rules are as follows. The quantitative score in the interval of 80 to 100 corresponds to the excellent level, and the teaching ability quantitative index is equal to × 100 × 1.10. The natural science is based on the distribution interval of the quantitative index of the excellent teachers in the sample, which is 90 to 100, which conforms to the right tail feature of the normal distribution, and the 1.10 times mapping is based on the statistical calibration result of the interval sample. The quantitative score in the interval of 60 to 79 corresponds to the good level, and the teaching ability quantitative index is equal to × 100 × 1.05. The natural science is based on the distribution interval of the quantitative index of the good teachers in the sample, which is 70 to 89, corresponding to the high-density area in the middle of the normal distribution, and the 1.05 times mapping fits the ability gradient difference of the teachers in this level. The quantitative score in the interval of 40 to 59 corresponds to the qualified level, and the teaching ability quantitative index is equal to × 100 × 0.95. The natural science is based on the distribution interval of the quantitative index of the qualified teachers in the sample, which is 50 to 69, which conforms to the left tail transition area of the normal distribution, and the 0.95 times mapping corrects the system deviation of the interval. The quantitative score below 40 corresponds to the unqualified level, and the teaching ability quantitative index is equal to × 100 × 0.80. The natural science is based on the distribution interval of the quantitative index of the unqualified teachers in the sample, which is 0 to 49, corresponding to the left tail low-density area of the normal distribution, and the 0.80 times mapping highlights the quantitative difference of the short board ability.

[0133] The quantitative score of the teacher A is 68.6, which belongs to the good level, and the corresponding teaching ability quantitative index is equal to 68.6*1.05, approximately equal to 72.0, and finally the teaching ability quantitative index of the teacher A is determined as 72, which accurately reflects the comprehensive teaching ability level of the experimental operation demonstration, interactive guidance and language expression of the teacher A.

[0134] The weight coefficient is not set by experience, but is realized by combining the analytic hierarchy process with mathematical testing. The scoring results of the evaluation group need to pass the consistency test to ensure logical consistency, and the final weight coefficient is calculated by the eigenvalue decomposition method, completely following the mathematical operation law and avoiding subjective speculation.

[0135] The quantitative score calculation process integrates various natural science models, and the Sigmoid activation function uses the nonlinear mapping characteristics to normalize the original data, eliminate extreme value interference, and conform to the mathematical law of data processing; the information entropy Based on the principle of probability statistics, the uniformity of the feature distribution is reflected, The value range of the adjustment coefficient is calibrated through a large amount of experimental data to ensure the stability of the calculation results.

[0136] The grade threshold of the teaching ability evaluation standard is determined based on the statistical law of big data, and by analyzing the quantitative results of more than 1000 VR teacher training samples, it is found that they conform to the normal distribution characteristics, and the grade threshold is equal to the mean value plus or minus 1 standard deviation, so that the evaluation standard has objective significance in a statistical sense, rather than simply being divided by man.

[0137] The beneficial effects are that through the above detailed implementation process, the node attribute parameters can be accurately collected from the teaching behavior feature map and classified by semantic association, the dimension grouping data is formed and converted into a dimension feature vector, and then the quantitative score is obtained by combining the preset weight configuration and the standard calculation, and finally the teaching ability quantitative index is mapped, each step in the whole process is closely connected and the operation is clear, which not only fully utilizes the multi-dimensional information of the teaching behavior characteristics, but also ensures the accuracy of the teaching ability quantitative index through reasonable calculation and mapping, and at the same time, the consideration of historical mean value, information entropy and other factors in the quantitative score calculation makes the obtained teaching ability quantitative index not only reflect the current teaching ability level of the teacher, but also reflect the stability and historical comparison of the ability performance, providing a scientific and reliable quantitative basis for the evaluation of the teaching ability of the teachers in the VR teacher training in colleges and universities.

[0138] S4、analyzing the dynamic evolution law of the teaching ability quantitative index to obtain the change trend of the teaching ability quantitative index;

[0139] In the embodiment of the present application, the analysis of the dynamic evolution law of the teaching ability quantitative index to obtain the change trend of the teaching ability quantitative index comprises:

[0140] serializing the teaching ability quantitative indicators to obtain a time sequence of the teaching ability quantitative indicators;

[0141]

[0142] According to the trend mode, the teaching ability quantitative indicators are synthesized in terms of trends to obtain a change trend of the teaching ability quantitative indicators.

[0143] When the teaching ability quantitative indicators are serially organized to obtain the time sequence of the teaching ability quantitative indicators, first, the division basis of the time dimension is determined, and the training period is divided into a plurality of continuous time units according to the actual development rhythm of the VR teacher training in colleges and universities, which can be specifically set as one time unit for each VR simulation teaching training class, or one time unit for completing the training of each teaching module, to ensure that the division of the time unit can completely cover the entire training process and has consistency. Then, the teacher teaching ability quantitative indicators generated after the end of each time unit are extracted from the teaching ability evaluation data storage library, which include specific numerical values of multiple dimensions such as communication ability, classroom control ability, and VR teaching tool application ability. Then, according to the order of the time units, the teaching ability quantitative indicator values of each dimension are arranged in turn, and each value is additionally marked with the corresponding time unit identifier, such as “the first VR simulation class-classroom control ability indicator 78”, “the second VR simulation class-classroom control ability indicator 82”, “the first teaching module-VR teaching tool application ability indicator 80”, etc. Finally, the time sequence of the teaching ability quantitative indicators is formed, which contains multi-dimensional indicator values and time identifiers in the order of time, to ensure that the sequence can completely reflect the specific situation of the teaching ability quantitative indicators of each dimension at different time nodes.

[0144] ​When the trend mode is analyzed by dynamic mode analysis on the time series of the index, the time series of the index is first split by dimension, and the analysis is carried out separately for the index series of each dimension to avoid interference between the change rules of indexes of different dimensions. In the analysis process, the index values of adjacent time units are compared one by one, the difference between adjacent values is calculated, and it is judged whether the value is rising, falling or stable, and at the same time the magnitude of each change is recorded, for example, a certain dimension index changes from 75 in the first time unit to 78 in the second time unit, the difference is +3, and it is judged as a small increase; from 78 in the second time unit to 85 in the third time unit, the difference is +7, and it is judged as a large increase. Subsequently, according to the change situation of continuous multiple time units, the overall change rule is summarized, if the index value in the continuous 5 time units is in the upward trend, and the upward amplitude is between 3-8 each time, there is no obvious downward or stable stage, then the trend mode of the time series of the index of this dimension is judged as "continuous upward mode"; if a certain dimension index slowly rises in 3 time units, then remains stable in 2 time units, and then slowly falls in 3 time units, and the rising and falling amplitudes are less than 5, then it is judged as "first rise, then stable, then fall mode"; if the index value fluctuates around a certain center value throughout the time period, the fluctuation amplitude is always less than 3, and there is no obvious continuous upward or downward trend, then it is judged as "smooth fluctuation mode", through such analysis process, the trend mode corresponding to each dimension time series of the index is finally obtained.

[0145] When synthesizing trends from quantitative indicators of teaching ability based on trend patterns, the first step is to identify the key time points of indicator changes under each trend pattern. These points include trend start points, trend turning points, and trend peak points. For example, in the "rise-stabilize-fall pattern," the starting point of the rising phase, the turning point from rising to stabilizing, the turning point from stabilizing to falling, and the ending point of the falling phase are all key points. Next, in a coordinate system with time as the horizontal axis and the quantitative indicator values ​​of teaching ability as the vertical axis, the time and indicator values ​​corresponding to each key point are converted into coordinate points and marked. For example, if a key point corresponds to "3rd VR simulation lesson - indicator 85," then the coordinate point is marked at the intersection of the horizontal axis "3rd" and the vertical axis "85." Then, based on the changing patterns of the trend pattern, these coordinate points are connected sequentially with smooth lines. During the connection process, it is necessary to ensure that the lines accurately reflect the changing trends between adjacent nodes. For example, the lines in the rising phase should show an upward sloping trend, the lines in the stabilizing phase should remain horizontal, and the lines in the falling phase should show a downward sloping trend. Meanwhile, next to the generated trend lines, the trend pattern name and key change information corresponding to the dimension indicator are marked, such as "continuous upward pattern, from the 1st to the 6th simulation class, the indicator rose from 72 to 91, with a cumulative increase of 19". Through this synthesis process, the change trend of the quantitative indicators of teaching ability in each dimension is finally obtained, and the dynamic evolution of the indicators over time is fully presented.

[0146] The beneficial effects are that, through the detailed implementation process described above, scattered quantitative indicators of teaching ability can be transformed into an ordered and structured time-series indicator sequence, accurately analyzing the specific trend patterns of each dimension of indicators and synthesizing intuitive and clear trends. This not only fully preserves the details of indicator changes at different time points but also clearly presents the overall evolutionary pattern of the indicators. This provides a comprehensive and accurate trend basis for subsequent analysis of teachers' teaching ability development stages and fitting development trajectories, ensuring that the dynamic evaluation of teachers' teaching ability during VR teacher training is more scientific and targeted, and helping to accurately grasp the improvement of teachers' teaching ability.

[0147] S5. Based on the key characteristics of the changing trend, perform stage trajectory fitting on the quantitative indicators of teaching ability to obtain the teacher's teaching ability development trajectory.

[0148] In this embodiment of the invention, the step of fitting the quantitative indicators of teaching ability to a stage trajectory based on the key characteristics of the changing trend to obtain the teacher's teaching ability development trajectory includes:

[0149] Extract the key features of the changing trend;

[0150] According to the time distribution and feature intensity of the key features, the teaching ability quantitative indicators are sequenced in states to obtain an evolution state sequence of the teaching ability quantitative indicators;

[0151] According to the logical association between states in the evolution state sequence, a connection relationship between the states is established.

[0152] Taking the states in the evolution state sequence as track nodes and the connection relationship as connection edges, a basic track framework of the teaching ability quantitative indicators is constructed.

[0153] The basic track framework is subjected to segmented track interpolation to obtain segmented track segments of the basic track framework.

[0154] The segmented track segments are subjected to transition zone smoothing to obtain the teaching ability development track of the teacher.

[0155] The segmented track segments are subjected to transition zone smoothing to obtain the teaching ability development track of the teacher, including:

[0156] According to the connection point features of the segmented track segments, a transition zone region of the segmented track segments is identified.

[0157] The track points in the transition zone region are subjected to curvature continuous optimization to obtain a smoothed track of the segmented track segments.

[0158] The smoothness of the smoothed track is verified to confirm the teaching ability development track of the teacher.

[0159] When extracting the key features of the change trend, the core elements in the teaching ability quantitative indicator change trend need to be comprehensively combed, including the starting time point and the ending time point in the index rising or falling process to determine the time range of the trend, the specific time node and the corresponding index value when the trend turns, the key position of the trend direction change, the highest value and the lowest value of the index in the trend to lock the extreme value features, the change amplitude of the index in unit time to determine the change rate features, and the integration of these extracted starting and ending points, turning nodes, extreme value data, and change rate to form the key features of the change trend.

[0160] When the evolution state sequence of the teaching ability quantitative index is obtained according to the time distribution and the feature intensity of the key features, first, all the key features are arranged in time sequence to determine the time distribution, and then the intensity level is divided according to the index change intensity reflected by the key features, and the state of the teaching ability quantitative index in each time interval is defined by combining the time distribution and the intensity level, for example, the index in a certain time interval is defined as a rapid promotion state if it presents rapid rise and high intensity level, and the index in a certain time interval is defined as a stable development state if it presents gentle change and low intensity level, and the states corresponding to all time intervals are arranged in time sequence to form the evolution state sequence of the teaching ability quantitative index.

[0161] When the connection relationship between the states is established according to the logical association between the states in the evolution state sequence, the attribute features of the adjacent two states in the evolution state sequence are analyzed one by one, the transition logic from the former state to the latter state is judged, if the former state is a rapid promotion state and the latter state is a stable development state, there is a logical association of “promotion and then tend to be stable” between the two states, then a one-way connection relationship from the rapid promotion state to the stable development state is established, if there is a mutual influence between the adjacent states, a bidirectional connection relationship is established, and the logical association basis of all adjacent state connection relationships needs to be clearly marked to ensure the rationality of the connection relationship.

[0162] When the basic trajectory framework of the teaching ability quantitative index is constructed by taking the states in the evolution state sequence as trajectory nodes and the connection relationship as connection edges, each state in the evolution state sequence is taken as an independent trajectory node, the time interval and the core feature corresponding to the state name are marked on the node, and then the connection relationship between the states established before is taken as the connection edge, and the trajectory nodes are connected in time sequence according to the time sequence of the evolution state sequence, forming a basic trajectory framework that can directly show the state evolution path of the teaching ability quantitative index, and the position of each node and the connection mode between the nodes need to be clearly presented in the framework.

[0163] When the segmented trajectory segment is obtained by segmenting and interpolating the basic trajectory framework, the entire framework is divided into several continuous segments according to the attribute change of the states in the basic trajectory framework, each segment contains a group of trajectory nodes and connection edges with similar change characteristics, and for each segment, the key time points and index values corresponding to the trajectory nodes in the segment are selected, and several intermediate points are supplemented between the adjacent two trajectory nodes, the intermediate points need to be reasonably set according to the index change law of the front and rear nodes to ensure that the index values of the intermediate points can accurately reflect the gradual change process between the nodes, and each segment forms an independent segmented trajectory segment after the intermediate points are supplemented.

[0164] When the transition zone smoothing is performed on the segmented trajectory segment to obtain the teaching ability development trajectory of the teacher, first, the transition zone region of the segmented trajectory segment is identified according to the connection point characteristics of the segmented trajectory segment, then the trajectory points in the transition zone region are optimized for curvature continuity to obtain a smoothed trajectory, and finally the smoothness of the smoothed trajectory is verified to confirm the teaching ability development trajectory.

[0165] When the transition zone region of the segmented trajectory segment is identified according to the connection point characteristics of the segmented trajectory segment, the trajectory trend and index change at the connection point of the adjacent two segmented trajectory segments are observed carefully, the region where the trajectory curvature changes obviously before and after the connection point is found, and the region is the transition zone region. The range of the transition zone region extends to the front and rear segmented trajectory segments with the connection point as the center, and the extension length is accurate to ensure the accuracy of the transition zone region.

[0166] When the trajectory points in the transition zone region are optimized for curvature continuity to obtain the smoothed trajectory of the segmented trajectory segment, the curvature values of each trajectory point in the transition zone region are analyzed one by one, the curvature difference of adjacent trajectory points is compared, the trajectory points with sudden change in curvature are adjusted, the curvature values of the sudden change points are modified to values that are connected with the curvature values of the front and rear trajectory points, the curvature of all trajectory points in the transition zone region presents a continuous and gradual change trend, and the adjacent segmented trajectory segments in the transition zone region can naturally connect to form a smoothed trajectory after the curvature optimization.

[0167] When the smoothness of the smoothed trajectory is verified to confirm the teaching ability development trajectory of the teacher, the curvature change rate between adjacent trajectory points along the smoothed trajectory is calculated in sequence, it is judged whether the curvature change rate is within a predetermined reasonable range, if the curvature change rates of all adjacent trajectory points are within the reasonable range and there is no obvious polyline or sudden change point in the entire trajectory, it is determined that the smoothed trajectory meets the smoothness requirement, the smoothed trajectory is confirmed as the teaching ability development trajectory of the teacher, and if there is a part that does not meet the smoothness requirement, the transition zone smoothing process is returned to be performed again until the requirement is met.

[0168] The beneficial effects are that through the above detailed and standardized implementation process, the change trend key features can be accurately extracted and the state sequence can be completed, a reasonable state connection relationship is established to construct a basic trajectory framework, a smoothed and accurate teaching ability development trajectory is obtained after the segmented interpolation and transition zone smoothing processing, the stage evolution path of the teaching ability of the teacher in the VR teacher training process is completely presented, comprehensive and reliable trajectory data support is provided for subsequent integration to generate comprehensive quantitative evaluation indexes, the evaluation of the teaching ability of the teacher is more dynamic and scientific, and the development law of the teaching ability of the teacher is accurately grasped.

[0169] S6, integrating the teaching ability quantitative index and the teaching ability development trajectory to generate the comprehensive quantitative evaluation index of the teacher.

[0170] In the embodiment of the present application, the integrated teaching ability quantitative index and the teaching ability development trajectory generate the comprehensive quantitative evaluation index of the teacher, which includes:

[0171] According to the target demand of the teaching ability evaluation standard, the weight coefficients of the teaching ability quantitative index and the teaching ability development trajectory are configured;

[0172] Based on the weight coefficients, the teaching ability quantitative index and the teaching ability development trajectory are fused to obtain the preliminary comprehensive index of the teacher;

[0173] Eliminate the dimension effect of the preliminary comprehensive index to obtain the comprehensive quantitative evaluation index of the teacher.

[0174] When configuring the weight coefficients of the teaching ability quantitative index and the teaching ability development trajectory according to the target demand of the teaching ability evaluation standard, first, the core target of each evaluation dimension in the teaching ability evaluation standard is determined, such as if the current teaching ability level of the teacher is the core target of the evaluation standard, the weight proportion of the teaching ability quantitative index is increased, and if the long-term improvement potential of the teaching ability of the teacher is the core target, the weight proportion of the teaching ability development trajectory is increased. Then, an evaluation group composed of university teaching experts, VR technology application experts and teacher training managers is established, and the importance of the teaching ability quantitative index and the teaching ability development trajectory is scored through collective discussion combined with the target demand of the evaluation standard. The average value of the scoring results is taken to determine the weight coefficients of the two in percentage form, ensuring that the sum of the weight coefficients is 100% and accurately matching the target orientation of the evaluation standard.

[0175] When the teaching ability quantitative index and the teaching ability development trajectory are fused based on the weight coefficients to obtain the preliminary comprehensive index of the teacher, first, the specific numerical value of the teaching ability quantitative index and the quantitative score corresponding to the teaching ability development trajectory are obtained, wherein the quantitative score of the teaching ability development trajectory needs to be converted into a specific numerical value according to the characteristics of the ability improvement range and stability reflected by the trajectory, and the pre-set trajectory scoring rules are referred to. Then, according to the configured weight coefficients, the numerical value of the teaching ability quantitative index is multiplied by the corresponding weight coefficient, and the quantitative score of the teaching ability development trajectory is multiplied by the corresponding weight coefficient. Then, the two product results are added, and the total sum is the preliminary comprehensive index of the teacher. During the calculation process, the accuracy of numerical calculation needs to be ensured to avoid affecting the reliability of the preliminary comprehensive index due to calculation errors.

[0176] When the dimension influence of the preliminary comprehensive index is eliminated to obtain the comprehensive quantitative evaluation index of the teacher, the preliminary comprehensive index data of all teachers participating in the VR teacher training is collected, and the maximum value and the minimum value in the data are counted. Then, the preliminary comprehensive index of each teacher is standardized according to a fixed formula, and the standardized value is obtained by subtracting the minimum value of all preliminary comprehensive index data from the preliminary comprehensive index value of the teacher and then dividing by the difference between the maximum value and the minimum value of all preliminary comprehensive index data. Finally, the standardized value is mapped and converted according to a preset scoring interval, and the converted value is the comprehensive quantitative evaluation index after eliminating the dimension influence, ensuring that the comprehensive quantitative evaluation indexes of different teachers are in the same dimension system and have the feasibility of horizontal comparison.

[0177] The beneficial effect is that by the above steps, the weight coefficient can be accurately configured according to the evaluation standard target, the scientific fusion of the teaching ability quantitative index and the development track is realized, and after the dimension elimination processing, the comprehensive quantitative evaluation index with a unified comparison standard is obtained, which not only reflects the current teaching ability level of the teacher, but also reflects the ability development potential, provides comprehensive, objective and comparable results for the evaluation of the VR teacher training effect in colleges and universities, and helps to accurately judge the training effectiveness and optimize the training scheme.

[0178] As shown in Figure 2 , it is a functional module diagram of a college VR teacher training simulation teaching data quantitative evaluation system provided by an embodiment of the present application.

[0179] The college VR teacher training simulation teaching data quantitative evaluation system 100 can be installed in an electronic device. According to the realized functions, the college VR teacher training simulation teaching data quantitative evaluation system 100 can include a multi-modal feature fusion module 101, a semantic network reconstruction module 102, a feature dimension mapping module 103, a dynamic evolution analysis module 104, a stage trajectory fitting module 105, and a comprehensive evaluation index generation module 106. The modules of the present application can also be referred to as units, which refer to a series of computer program segments that can be executed by an electronic device processor and can complete a fixed function, stored in the memory of the electronic device.

[0180] In this embodiment, the functions of each module / unit are as follows:

[0181] The multi-modal feature fusion module 101 is configured to perform multi-modal feature fusion on the teaching behavior data of the teacher to obtain teaching behavior features of the teaching behavior data.

[0182] The semantic network reconstruction module 102 is configured to perform semantic network reconstruction on the teaching behavior features to obtain a teaching behavior feature map of the teaching behavior features.

[0183] The feature dimension mapping module 103 is configured to perform feature dimension mapping on the teaching behavior feature graph according to the node attribute of the teaching behavior feature graph, so as to obtain the teaching ability quantitative index of the teacher.

[0184] The dynamic evolution analysis module 104 is configured to analyze the dynamic evolution law of the teaching ability quantitative index, so as to obtain the change trend of the teaching ability quantitative index.

[0185] The stage trajectory fitting module 105 is configured to perform stage trajectory fitting on the teaching ability quantitative index according to the key feature of the change trend, so as to obtain the teaching ability development trajectory of the teacher.

[0186] The comprehensive evaluation index generation module 106 is configured to integrate the teaching ability quantitative index and the teaching ability development trajectory, so as to generate the comprehensive quantitative evaluation index of the teacher.

[0187] In several embodiments provided in the present application, it should be understood that the disclosed method and system can be implemented in other ways. For example, the system embodiments described above are only illustrative, for example, the division of the modules is only a logical function division, and another division mode can be used in actual implementation.

[0188] The modules described as separate components can or can not be physically separated, and the components displayed as modules can or can not be physical units, that is, they can be located in one place or distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiment according to actual needs.

[0189] In addition, each functional module in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of hardware plus software function modules.

[0190] It is obvious for those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and the present application can be implemented in other specific forms without departing from the spirit or essential characteristics of the present application.

[0191] The embodiment of the present application can acquire and process related data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology and application system for using digital computers or digital computer controlled machines to simulate, extend and expand human intelligence, perceive environment, acquire knowledge and use knowledge to obtain optimal results.

[0192] Finally, it should be noted that the above examples are merely intended to illustrate the technical solutions of the present application and not to limit the present application. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalently replaced without departing from the spirit and scope of the present application.

Claims

1. A high school VR teacher training simulation teaching data quantification evaluation method, characterized in that, The method comprises: S1, multi-modal feature fusion is carried out on the teaching behavior data of the teacher, and teaching behavior features of the teaching behavior data are obtained; S2, semantic network reconstruction is carried out on the teaching behavior features, and a teaching behavior feature map of the teaching behavior features is obtained; S3, according to the node attribute of the teaching behavior feature map, feature dimension mapping is carried out on the teaching behavior feature map, and a teaching ability quantitative index of the teacher is obtained; S4, the dynamic evolution law of the teaching ability quantitative index is analyzed, and a change trend of the teaching ability quantitative index is obtained; S5, according to the key features of the change trend, stage trajectory fitting is carried out on the teaching ability quantitative index, and a teaching ability development trajectory of the teacher is obtained; S6, the teaching ability quantitative index and the teaching ability development trajectory are integrated, and a comprehensive quantitative evaluation index of the teacher is generated.

2. The high school VR instructor training simulation teaching data quantitative evaluation method of claim 1, wherein, The multi-modal feature fusion on the teaching behavior data of the teacher to obtain the teaching behavior features of the teaching behavior data comprises: obtaining the teaching behavior data of the teacher; extracting the speech features of the original speech waveform in the teaching behavior data; tracking the skeleton key point data of the teaching behavior data to obtain the posture features of the teacher; extracting the features of the interaction frequency, response time and interaction depth of the teaching behavior data to obtain the interaction features of the teacher; performing semantic alignment on the speech features, the posture features and the interaction features to obtain the preliminary teaching behavior features of the teacher; eliminating the redundant features of the preliminary teaching behavior features to obtain the teaching behavior features of the teaching behavior data.

3. The high school VR instructor training simulation teaching data quantitative evaluation method of claim 1, wherein, The semantic network reconstruction on the teaching behavior features to obtain the teaching behavior feature map of the teaching behavior features comprises: performing semantic correlation analysis on the teaching behavior features to generate a feature similarity matrix of the teaching behavior features; according to the correlation strength in the feature similarity matrix, when the correlation strength reaches a preset threshold, a connection relationship is established between the related features of the teaching behavior features; taking the related features as nodes and the connection relationship as connection edges, an initial feature map of the teaching behavior features is constructed; performing semantic relationship enhancement on the initial feature map to obtain an optimized feature map of the initial feature map; verifying the connectivity of the optimized feature map to confirm the teaching behavior feature map of the teaching behavior features.

4. The high school VR instructor training simulation teaching data quantitative evaluation method of claim 3, wherein, The semantic correlation analysis on the teaching behavior features to generate the feature similarity matrix of the teaching behavior features comprises: numerical value processing is performed on the teaching behavior features, and semantic information of the teaching behavior features is retained to obtain a feature vector of the teaching behavior features; according to the distribution characteristics of the feature vector, a semantic space framework of the feature vector is constructed; in the semantic space framework, the relative position and distribution mode of the feature vector are analyzed to obtain similarity relationship data of the feature vector; the similarity relationship data is organized in a matrix structure to obtain the feature similarity matrix of the teaching behavior features.

5. The high school VR instructor training simulation teaching data quantitative evaluation method of claim 1, wherein, The node attribute of the teaching behavior feature map is mapped to a feature dimension to obtain a teaching ability quantitative index of the teacher, including: Collecting node attribute parameters of the teaching behavior feature map; According to the semantic correlation of the node attribute parameters, the node attribute parameters are classified according to the semantic correlation to obtain dimension grouping data of the node attribute parameters; The dimension grouping data is mapped to a vector space to obtain a dimension feature vector of the dimension grouping data; According to a preset feature weight configuration, a quantitative score of the dimension feature vector is calculated, wherein the calculation formula of the quantitative score is as follows: ； In the formula, This represents the quantized score of the feature vector of the stated dimension. Indicates the first The weight coefficients of the feature vectors in each dimension. The dimensional feature vector represents the first... Each component value This represents the Sigmoid activation function. Represents the dimensional feature vector of the th dimension Historical mean of each component, This represents the preset information entropy adjustment coefficient. Represents the dimensional feature vector Information entropy This represents the number of dimensions in the dimensional feature vector. This represents the summation operation. This represents the square root operation; The quantitative score is mapped to a preset teaching ability evaluation standard to obtain the teaching ability quantitative index of the teacher.

6. The high school VR instructor training simulation teaching data quantitative evaluation method of claim 1, wherein, The dynamic evolution law of the teaching ability quantitative index is analyzed to obtain the change trend of the teaching ability quantitative index, including: The teaching ability quantitative index is serialized to obtain a time sequence index sequence of the teaching ability quantitative index; The time sequence index sequence is dynamically patterned to obtain a trend mode of the time sequence index sequence; According to the trend mode, the teaching ability quantitative index is trend synthesized to obtain the change trend of the teaching ability quantitative index.

7. The high school VR instructor training simulation teaching data quantitative evaluation method of claim 1, wherein, According to the key features of the change trend, the teaching ability quantitative index is fitted to obtain the teaching ability development trajectory of the teacher, including: Extracting key features of the change trend; According to the time distribution and feature intensity of the key features, the teaching ability quantitative index is state-serialized to obtain an evolution state sequence of the teaching ability quantitative index; According to the logical association between states in the evolution state sequence, a connection relationship between the states is established; The states in the evolution state sequence are taken as trajectory nodes, and the connection relationship is taken as a connection edge to construct a basic trajectory framework of the teaching ability quantitative index; The basic trajectory framework is segmented and interpolated to obtain a segmented trajectory segment of the basic trajectory framework; The transition zone of the segmented trajectory segment is smoothed to obtain the teaching ability development trajectory of the teacher.

8. The high school VR instructor training simulation teaching data quantitative evaluation method of claim 7, wherein, The transition zone of the segmented trajectory segment is smoothed to obtain the teaching ability development trajectory of the teacher, including: According to the connection point features of the segmented trajectory segment, the transition zone region of the segmented trajectory segment is identified; The curvature of the trajectory points in the transition zone region is continuously optimized to obtain a smoothed trajectory of the segmented trajectory segment; The smoothness of the smoothed trajectory is verified to confirm the teaching ability development trajectory of the teacher.

9. The high school VR instructor training simulation teaching data quantitative evaluation method of claim 5, wherein, The teaching ability quantitative index and the teaching ability development trajectory are integrated to generate a comprehensive quantitative evaluation index of the teacher, including: According to the target demand of the teaching ability evaluation standard, the weight coefficients of the teaching ability quantitative index and the teaching ability development trajectory are configured; Based on the weight coefficients, the teaching ability quantitative index and the teaching ability development trajectory are fused to obtain a preliminary comprehensive index of the teacher; The dimension effect of the preliminary comprehensive index is eliminated to obtain the comprehensive quantitative evaluation index of the teacher.

10. A high school VR teacher training simulation teaching data quantitative evaluation system, characterized in that, The system comprises: The multi-modal feature fusion module is configured to perform multi-modal feature fusion on the teaching behavior data of the teacher to obtain teaching behavior features of the teaching behavior data. The semantic network reconstruction module is configured to perform semantic network reconstruction on the teaching behavior features to obtain a teaching behavior feature map of the teaching behavior features. The feature dimension mapping module is configured to perform feature dimension mapping on the teaching behavior feature map according to node attributes of the teaching behavior feature map to obtain a teaching ability quantitative index of the teacher. The dynamic evolution analysis module is configured to analyze a dynamic evolution law of the teaching ability quantitative index to obtain a change trend of the teaching ability quantitative index. The stage trajectory fitting module is configured to perform stage trajectory fitting on the teaching ability quantitative index according to key features of the change trend to obtain a teaching ability development trajectory of the teacher. The comprehensive evaluation index generation module is configured to integrate the teaching ability quantitative index and the teaching ability development trajectory to generate a comprehensive quantitative evaluation index of the teacher.

Images