A large-scale AI model training method and system for vocational school teaching

Abstract

The present application provides a method and system for training a large AI model for teaching in vocational schools, which obtains historical interaction data when a terminal device trains knowledge scenario teaching resources in a preset knowledge scenario teaching resource library; performs feature extraction based on the historical interaction data to obtain teacher feature information corresponding to the historical interaction data; obtains a training adjustment coefficient based on the teacher interaction features; generates a mind map related to the knowledge scenario teaching resources, and extracts mind map features from the mind map; inputs the teacher interaction features, teacher knowledge scenario teaching features, mind map features and training adjustment coefficients into a preset intelligent training model for training to obtain an AI large model; the above method can not only better meet the personalized adaptation capabilities of teaching resources, help improve the pertinence of the AI ​​large model to the teaching scenarios of vocational schools, but also effectively optimize training efficiency, reduce invalid training data, increase model convergence speed, and improve resource utilization.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and specifically to a method and system for training a large AI model for teaching in vocational schools. Background Art

[0002] In vocational colleges, existing large AI models are typically trained based on general teaching data and lack adaptation to teachers' individual learning characteristics (such as interaction habits, teaching scenarios, and so on). This results in the model's output of teaching courses not being well-matched with teachers' actual needs, making it difficult to meet the precision requirements of current vocational college teaching reforms. Furthermore, traditional training methods fail to fully incorporate the dynamic adjustment needs of historical user interaction data, resulting in inefficient model training.

[0003] Therefore, developing a large-scale AI model training method and system for vocational school teaching has important practical significance and broad application prospects. Summary of the Invention

[0004] The purpose of this application is to provide a large-scale AI model training method and system for vocational school teaching, which can not only improve the matching degree between teaching courses and teachers' actual needs, better meet the personalized adaptation capabilities of teaching resources, but also improve model training efficiency.

[0005] In a first aspect, the present application provides a method for training a large AI model for teaching in vocational schools, comprising the following steps:

[0006] Obtain historical interaction data when the terminal device is training the knowledge scenario teaching resources in the preset knowledge scenario teaching resource library;

[0007] Extracting features from historical interaction data to obtain teacher feature information corresponding to the historical interaction data; wherein the teacher feature information includes teacher interaction features and teacher knowledge scenario teaching features;

[0008] Obtaining training adjustment coefficients based on teacher interaction characteristics;

[0009] Generate a mind map related to knowledge scenario teaching resources and extract mind map features from the mind map;

[0010] Input teacher interaction features, teacher knowledge scenario teaching features, mind map features, and training adjustment coefficients into the preset intelligent training model for training to obtain an AI large model;

[0011] Among them, the preset intelligent training model includes a feature data processing module, a teacher knowledge scenario teaching feature calculation module, a mind map feature calculation module, a teacher interaction feature calculation module, a comprehensive analysis module and a training result output module;

[0012] The feature data processing module is used to perform feature extraction and data standardization processing on the teacher interaction feature to obtain a first feature standardized data value of the teacher interaction feature and a second feature standardized data value of the teacher interaction feature;

[0013] The teacher knowledge scenario teaching feature calculation module is used to perform feature extraction and data standardization processing on the teacher knowledge scenario teaching feature to obtain the third feature standardized data value of the teacher knowledge scenario teaching feature;

[0014] The mind map feature calculation module includes a sample representation component, a time representation component, and a splicing component; the sample representation component is used to extract node features, edge features, and structural features; the time representation component is used to extract time series features and time-dependent features from historical interaction data; the splicing component is used to splice the node features, edge features, and structural features extracted by the sample representation component with the time series features and time-dependent features extracted by the time representation component to form a feature vector, and use the feature vector as the merged data value;

[0015] The teacher interaction feature calculation module is used to calculate the comprehensive feature training value of the teacher interaction feature based on the first feature standardized data value and the second feature standardized data value;

[0016] The comprehensive analysis module is used to comprehensively process the third feature standardized data value, the comprehensive feature training value and the merged data value to obtain the optimal training data value;

[0017] The training result output module is used to obtain the preset AI large model training data information based on the optimal training data value and train the AI ​​large model.

[0018] Optionally, extracting features from the historical interaction data to obtain teacher feature information corresponding to the historical interaction data specifically includes the following steps:

[0019] Performing data classification on the historical interaction data to obtain an interaction information set corresponding to the historical interaction data; wherein the interaction information set includes teacher interaction information and teacher knowledge scenario teaching feature set;

[0020] Extract features of teacher interaction information in the interaction information set to obtain teacher interaction features corresponding to historical interaction data;

[0021] Feature extraction is performed on the teacher knowledge scenario teaching feature set in the interactive information set to obtain the teacher knowledge scenario teaching feature set corresponding to the historical interactive data.

[0022] Optionally, obtaining the training adjustment coefficient according to the teacher interaction characteristics specifically includes the following steps:

[0023] Obtaining the interaction standard matching coefficient corresponding to the teacher interaction feature according to the teacher interaction feature;

[0024] Obtain the comprehensive interaction matching degree corresponding to the teacher's interaction characteristics according to the interaction standard matching coefficient corresponding to the teacher's interaction characteristics;

[0025] Compare the comprehensive interaction matching degree corresponding to the teacher's interaction feature with the interaction threshold. If the comprehensive interaction matching degree corresponding to the teacher's interaction feature is less than the interaction threshold, generate an interaction adjustment correction request corresponding to the teacher's interaction feature.

[0026] The training adjustment coefficient is calculated based on the interactive adjustment correction request.

[0027] Optionally, calculating the training adjustment coefficient according to the interactive adjustment correction request specifically includes the following steps:

[0028] Identifying the interaction adjustment and correction request and obtaining the terminal device interaction characteristics corresponding to the interaction adjustment and correction request; wherein the terminal device interaction characteristics include terminal device interaction habits and / or terminal device interaction time;

[0029] The terminal interaction feature recognition degree corresponding to the interactive adjustment correction request is calculated based on the terminal device interaction feature corresponding to the interactive adjustment correction request;

[0030] Comparing the terminal interaction feature recognition degree with the interaction adjustment interval, and generating a first adjustment instruction if the terminal interaction feature recognition degree is within a first preset adjustment interval;

[0031] If the terminal interaction feature recognition degree is within a second preset adjustment range, generating a second adjustment instruction;

[0032] If the terminal feature recognition degree exceeds the second preset adjustment range, a third adjustment instruction is generated; wherein the first adjustment instruction means no adjustment processing; the second adjustment instruction means partial adjustment processing; and the third adjustment instruction means full adjustment processing;

[0033] The training adjustment coefficient is obtained by calculating the adjustment degrees corresponding to the first adjustment instruction, the second adjustment instruction, and the third adjustment instruction.

[0034] Optionally, the comprehensive analysis module performs comprehensive processing on the third feature normalized data value, the comprehensive feature training value and the merged data value to obtain the optimal training data value, specifically comprising the following steps:

[0035] Obtain the knowledge scenario teaching feature weight corresponding to the teacher's knowledge scenario teaching feature;

[0036] The knowledge scenario teaching feature training value corresponding to the teacher's knowledge scenario teaching feature is calculated based on the knowledge scenario teaching feature weight and the third feature standardized data value;

[0037] Obtain the adjustment coefficient weight corresponding to the teacher interaction characteristics;

[0038] The teacher interaction feature training value corresponding to the teacher interaction feature is obtained according to the adjustment coefficient weight and the comprehensive feature training value;

[0039] The optimal training data value is obtained by comprehensive calculation based on the knowledge scenario teaching feature training value, teacher interaction feature training value and merged data value.

[0040] Optionally, the step of extracting features of the teacher interaction information in the interaction information set to obtain teacher interaction features corresponding to the historical interaction data specifically includes the following steps:

[0041] Data classification is performed on the interactive information set to obtain teacher interactive information and a knowledge scenario teaching feature set; wherein the knowledge scenario teaching feature set refers to teacher learning feature information corresponding to the knowledge scenario features under the teaching scenario;

[0042] Converting the learning features corresponding to the knowledge scenario features in the knowledge scenario teaching feature set into the knowledge scenario teaching feature learning values;

[0043] The teacher interaction information is matched with the learning values ​​of the teaching features of the learning scenario to obtain the teacher interaction features.

[0044] Optionally, extracting the teacher's knowledge scenario teaching features to obtain a set of teacher's knowledge scenario teaching features corresponding to the historical interaction data specifically includes the following steps:

[0045] Obtain the number of knowledge scenario features in the knowledge scenario teaching feature set and the number of times the knowledge scenario features are learned and used;

[0046] The comprehensive learning degree of the knowledge scenario features is calculated based on the number of knowledge scenario features and the number of times the knowledge scenario features are learned and used;

[0047] If the comprehensive learning degree of the knowledge scenario feature meets the preset learning degree, the learning scenario teaching feature set corresponding to the knowledge scenario feature is obtained, and the teacher knowledge scenario teaching feature set is generated.

[0048] Optionally, the method further includes:

[0049] Teachers use the VR scene editor to obtain the created virtual teaching scene data; the virtual teaching scene data includes three-dimensional space coordinate data, teaching props layout data and dynamic interaction event configuration data;

[0050] Obtain the number of props, interaction node density, and spatial segmentation dimension, and calculate the scene complexity coefficient based on the number of props, interaction node density, and spatial segmentation dimension;

[0051] Dynamically couple the scene complexity coefficient with the training adjustment coefficient to generate the environmental fitness parameter;

[0052] The teacher interaction characteristics, teacher knowledge scenario teaching characteristics, mind map characteristics and environmental adaptability parameters are input into the preset intelligent training model for training to obtain the AI ​​large model.

[0053] Optionally, the method further includes:

[0054] Construct virtual-reality fusion training verification module:

[0055] Embed knowledge graph anchors in VR scenes, triggering multimodal explanations of related knowledge points when students interact with specific teaching props;

[0056] Collect students' attention distribution data through eye tracking and gesture recognition;

[0057] Compare the real-time attention heat map with the preset knowledge importance distribution map to generate scenario teaching effectiveness evaluation indicators and feed them back to the AI ​​big model.

[0058] In a second aspect, the present application provides a large-scale AI model training system for vocational school teaching, including:

[0059] The first acquisition unit is used to acquire historical interaction data when the terminal device is training the knowledge scenario teaching resources in the preset knowledge scenario teaching resource library;

[0060] An extraction unit is configured to extract features from the historical interaction data to obtain teacher feature information corresponding to the historical interaction data; wherein the teacher feature information includes teacher interaction features and teacher knowledge scenario teaching features;

[0061] A second acquisition unit is used to obtain a training adjustment coefficient according to the teacher interaction characteristics;

[0062] A generation unit, used to generate a mind map related to the knowledge scenario teaching resources and extract mind map features from the mind map;

[0063] The training unit inputs the teacher interaction characteristics, teacher knowledge scenario teaching characteristics, mind map characteristics and training adjustment coefficients into the preset intelligent training model to train and obtain the AI ​​large model;

[0064] Among them, the preset intelligent training model includes a feature data processing module, a teacher knowledge scenario teaching feature calculation module, a mind map feature calculation module, a teacher interaction feature calculation module, a comprehensive analysis module and a training result output module;

[0065] The feature data processing module is used to perform feature extraction and data standardization processing on the teacher interaction feature to obtain a first feature standardized data value of the teacher interaction feature and a second feature standardized data value of the teacher interaction feature;

[0066] The teacher knowledge scenario teaching feature calculation module is used to perform feature extraction and data standardization processing on the teacher knowledge scenario teaching feature to obtain the third feature standardized data value of the teacher knowledge scenario teaching feature;

[0067] The mind map feature calculation module includes a sample representation component, a time representation component, and a splicing component; the sample representation component is used to extract node features, edge features, and structural features; the time representation component is used to extract time series features and time-dependent features from historical interaction data; the splicing component is used to splice the node features, edge features, and structural features extracted by the sample representation component with the time series features and time-dependent features extracted by the time representation component to form a feature vector, and use the feature vector as the merged data value;

[0068] The teacher interaction feature calculation module is used to calculate the comprehensive feature training value of the teacher interaction feature based on the first feature standardized data value and the second feature standardized data value;

[0069] The comprehensive analysis module is used to comprehensively process the third feature standardized data value, the comprehensive feature training value and the merged data value to obtain the optimal training data value;

[0070] The training result output module is used to obtain the preset AI large model training data information based on the optimal training data value and train the AI ​​large model.

[0071] Compared with the existing technology, the present application provides a method for training a large AI model for teaching in vocational schools, which obtains historical interaction data when a terminal device trains knowledge scenario teaching resources in a preset knowledge scenario teaching resource library; extracts features based on the historical interaction data to obtain teacher feature information corresponding to the historical interaction data; wherein the teacher feature information includes teacher interaction features and teacher knowledge scenario teaching features; obtains a training adjustment coefficient based on the teacher interaction features; generates a mind map related to the knowledge scenario teaching resources, and extracts mind map features from the mind map; inputs the teacher interaction features, teacher knowledge scenario teaching features, mind map features, and training adjustment coefficients into a preset intelligent training model for training to obtain an AI large model; Among them, the preset intelligent training model includes a feature data processing module, a teacher knowledge scenario teaching feature calculation module, a mind map feature calculation module, a teacher interaction feature calculation module, a comprehensive analysis module and a training result output module; the feature data processing module is used to perform feature extraction and data standardization processing on the teacher interaction feature, and obtain the first feature standardized data value of the teacher interaction feature and the second feature standardized data value of the teacher interaction feature; the teacher knowledge scenario teaching feature calculation module is used to perform feature extraction and data standardization processing on the teacher knowledge scenario teaching feature, and obtain the third feature standardized data value of the teacher knowledge scenario teaching feature; the mind map feature calculation module includes a sample representation component, a time representation component and a splicing component; The sample characterization component is used to extract node features, edge features and structural features, the time characterization component is used to extract time series features and time-dependent features from historical interaction data, and the splicing component is used to splice the node features, edge features and structural features extracted by the sample characterization component with the time series features and time-dependent features extracted by the time characterization component to form a feature vector, and use the feature vector as the merged data value; the teacher interaction feature calculation module is used to calculate based on the first feature standardized data value and the second feature standardized data value to obtain the comprehensive feature training value of the teacher interaction feature; the comprehensive analysis module is used to perform comprehensive processing on the third feature standardized data value, the comprehensive feature training value and the merged data value to obtain the optimal training data value ; The training result output module is used to obtain the preset AI big model training data information according to the optimal training data value, and train the AI ​​big model; the above method can not only accurately predict and better adapt to the differences of each teacher user and dynamically adapt to the changing data conditions, improve the matching degree between teaching courses and teachers' actual needs, better meet the personalized adaptation capabilities of teaching resources, and help improve the pertinence of AI big models to vocational school teaching scenarios, but also effectively optimize training efficiency, based on the dynamic adjustment mechanism of interactive matching degree, reduce invalid training data, improve model convergence speed, and improve resource utilization through comprehensive learning degree screening of knowledge scenario characteristics, give priority to the use of high-frequency teaching resources, and optimize resource allocation. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present application. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0073] Figure 1 A flowchart of a large-scale AI model training method for vocational school teaching provided in this application is shown;

[0074] Figure 2 A schematic diagram of a large-scale AI model training system for vocational school teaching provided by this application is shown;

[0075] Figure 3 A schematic diagram of an electronic device provided by the present application is shown. DETAILED DESCRIPTION

[0076] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0077] It should be noted that, unless otherwise specified, the technical or scientific terms used in this application should have the common meanings understood by those skilled in the art to which this application belongs.

[0078] In addition, the terms "first" and "second" are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or elements is not limited to the listed steps or elements, but may optionally include steps or elements not listed, or may optionally include other steps or elements inherent to the process, method, product, or apparatus.

[0079] Please refer to Figure 1 , Figure 1 The embodiment of the present application provides a method for training a large AI model for teaching in vocational schools, which includes the following steps:

[0080] Step S101: Acquire historical interaction data when a terminal device is training a knowledge scenario teaching resource in a preset knowledge scenario teaching resource library;

[0081] Step S102: Extracting features from the historical interaction data to obtain teacher feature information corresponding to the historical interaction data; wherein the teacher feature information includes teacher interaction features (such as operating habits, teaching rhythm, software click frequency, error correction records, etc.) and teacher knowledge scenario teaching features (such as teaching content selection, teaching methods, and mastery of a certain course knowledge point, etc.);

[0082] Step S103: obtaining a training adjustment coefficient based on the teacher interaction characteristics;

[0083] Step S104: generating a mind map related to the knowledge scenario teaching resources, and extracting mind map features from the mind map;

[0084] Step S105: Input teacher interaction features, teacher knowledge scenario teaching features, mind map features, and training adjustment coefficients into a preset intelligent training model for training to obtain an AI large model;

[0085] Among them, the preset intelligent training model includes a feature data processing module, a teacher knowledge scenario teaching feature calculation module, a mind map feature calculation module, a teacher interaction feature calculation module, a comprehensive analysis module and a training result output module;

[0086] The feature data processing module is used to perform feature extraction and data standardization processing on the teacher interaction feature to obtain a first feature standardized data value of the teacher interaction feature and a second feature standardized data value of the teacher interaction feature;

[0087] The teacher knowledge scenario teaching feature calculation module is used to perform feature extraction and data standardization processing on the teacher knowledge scenario teaching feature to obtain the third feature standardized data value of the teacher knowledge scenario teaching feature;

[0088] The mind map feature calculation module includes a sample representation component, a time representation component, and a splicing component; the sample representation component is used to extract node features, edge features, and structural features; the time representation component is used to extract time series features and time-dependent features from historical interaction data; the splicing component is used to splice the node features, edge features, and structural features extracted by the sample representation component with the time series features and time-dependent features extracted by the time representation component to form a feature vector, and use the feature vector as the merged data value;

[0089] The teacher interaction feature calculation module is used to calculate the comprehensive feature training value of the teacher interaction feature based on the first feature standardized data value and the second feature standardized data value;

[0090] The comprehensive analysis module is used to comprehensively process the third feature standardized data value, the comprehensive feature training value and the merged data value to obtain the optimal training data value;

[0091] The training result output module is used to obtain the preset AI large model training data information based on the optimal training data value and train the AI ​​large model.

[0092] In step S102, extracting features from the historical interaction data to obtain teacher feature information corresponding to the historical interaction data specifically includes the following steps:

[0093] Step S1021: classify the historical interaction data to obtain an interaction information set corresponding to the historical interaction data; wherein the interaction information set includes teacher interaction information and teacher knowledge scenario teaching feature set;

[0094] Step S1022: extracting features of the teacher interaction information (such as interaction frequency, interaction mode, interaction duration, etc.) in the interaction information set to obtain teacher interaction features (such as interaction activity, interaction effect, etc.) corresponding to the historical interaction data;

[0095] Step S1023: extract features of the teacher knowledge scenario teaching feature set (such as teaching methods, teaching content, teaching effects, etc.) in the interactive information set to obtain the teacher knowledge scenario teaching feature set (such as teaching professionalism, teaching innovation, etc.) corresponding to the historical interactive data.

[0096] Through the above three steps, teacher-related feature information can be extracted from historical interaction data, including teacher interaction features and teacher knowledge scenario teaching features. These feature information can be used for subsequent teacher evaluation, teaching improvement and other aspects.

[0097] In step S103, obtaining the training adjustment coefficient according to the teacher interaction characteristics specifically includes the following steps:

[0098] Obtaining the interaction standard matching coefficient corresponding to the teacher interaction feature according to the teacher interaction feature;

[0099] Obtaining the comprehensive interaction matching degree corresponding to the teacher interaction feature according to the interaction standard matching coefficient corresponding to the teacher interaction feature, which can be calculated by weighted or average processing;

[0100] Compare the comprehensive interaction matching degree corresponding to the teacher's interaction feature with the interaction threshold. If the comprehensive interaction matching degree corresponding to the teacher's interaction feature is less than the interaction threshold, generate an interaction adjustment correction request corresponding to the teacher's interaction feature.

[0101] The training adjustment coefficient is calculated based on the interactive adjustment correction request.

[0102] Through the above steps, the teacher's interaction characteristics can be used to evaluate their match with the standard interaction behavior, and adjustment requests can be generated when necessary. These adjustment requests are further converted into training adjustment coefficients, which are used to optimize the teacher's teaching behavior or the system's training process to improve teaching effectiveness and interaction quality.

[0103] The step of calculating the training adjustment coefficient according to the interactive adjustment correction request specifically includes the following steps:

[0104] Identify the interactive adjustment correction request and obtain the terminal device interaction characteristics corresponding to the interactive adjustment correction request. The terminal device interaction characteristics include terminal device interaction habits and / or terminal device interaction time. For example, teachers may be more accustomed to using certain functions on the PC or tend to conduct teaching activities during specific time periods. This information is very important for providing personalized adjustment suggestions.

[0105] The terminal interaction feature recognition degree corresponding to the interactive adjustment correction request is calculated based on the terminal device interaction features corresponding to the interactive adjustment correction request. The terminal interaction feature recognition degree reflects the system's understanding of the teacher's interactive behavior. A high recognition degree means that the system can accurately predict the teacher's behavior pattern, which helps to provide more accurate adjustment suggestions.

[0106] Comparing the terminal interaction feature recognition degree with the interaction adjustment interval, and generating a first adjustment instruction if the terminal interaction feature recognition degree is within a first preset adjustment interval;

[0107] If the terminal interaction feature recognition degree is within a second preset adjustment range, generating a second adjustment instruction;

[0108] If the terminal feature recognition degree exceeds the second preset adjustment range, a third adjustment instruction is generated; wherein the first adjustment instruction means no adjustment processing; the second adjustment instruction means partial adjustment processing; and the third adjustment instruction means full adjustment processing;

[0109] The training adjustment coefficient is calculated based on the adjustment degrees corresponding to the first, second, and third adjustment instructions. The final training adjustment coefficient, calculated based on the adjustment instructions, will guide the teacher's subsequent training process. The coefficient size reflects the degree of adjustment required, ranging from no adjustment to full adjustment. This graded adjustment method ensures that the system can provide personalized training suggestions based on the teacher's specific situation, thereby improving training effectiveness.

[0110] Through this process, we can comprehensively analyze teachers' teaching performance, identify their strengths and weaknesses, and provide targeted training and adjustment suggestions. This approach not only improves teachers' teaching quality, but also enhances students' learning experience and promotes the development of educational technology.

[0111] In step S105, the comprehensive analysis module performs comprehensive processing on the third feature normalized data value, the comprehensive feature training value and the merged data value to obtain the optimal training data value, which specifically includes the following steps:

[0112] Obtain the knowledge scenario teaching feature weights corresponding to the teacher's knowledge scenario teaching features; knowledge scenario teaching features may include the teacher's explanation method, language expression, emotional transmission, etc. on different knowledge points; while interaction features may include the frequency of interaction between teachers and students, interaction quality, feedback mechanism, etc. The weights reflect the importance and influence of different knowledge scenarios in teaching;

[0113] The knowledge scenario teaching feature training value corresponding to the teacher's knowledge scenario teaching feature is calculated based on the knowledge scenario teaching feature weight and the third feature standardized data value;

[0114] Obtain the adjustment coefficient weights corresponding to the teacher interaction features; these weights reflect the adjustment degree and importance of the teacher interaction features in the training process;

[0115] The teacher interaction feature training value corresponding to the teacher interaction feature is calculated based on the adjustment coefficient weight and the comprehensive feature training value; this training value is a quantitative representation of the teacher's performance and adjustment needs during the interaction process;

[0116] Based on the training value of knowledge scenario teaching characteristics, the training value of teacher interaction characteristics and the merged data value, a comprehensive calculation is performed to obtain the optimal training data value. The optimal training data value is the result of comprehensive consideration of the teacher's knowledge scenario teaching characteristics, interaction characteristics and other relevant data, and is used to guide subsequent training and optimization processes.

[0117] The above process ensures the comprehensiveness and accuracy of training data, providing a solid data foundation for subsequent training and optimization. Through weight allocation, feature training value calculation, and comprehensive calculation, this method can more effectively capture and utilize the teacher's teaching characteristics and interaction characteristics, thereby improving training effectiveness and teaching quality.

[0118] In step S1022, feature extraction is performed on the teacher interaction information in the interaction information set to obtain teacher interaction features corresponding to the historical interaction data, specifically including the following steps:

[0119] Data classification is performed on the interactive information set to obtain teacher interactive information and a knowledge scenario teaching feature set; wherein the knowledge scenario teaching feature set refers to teacher learning feature information corresponding to the knowledge scenario features under the teaching scenario;

[0120] Converting the learning features corresponding to the knowledge scenario features in the knowledge scenario teaching feature set into the knowledge scenario teaching feature learning values;

[0121] The teacher interaction information is matched with the learning values ​​of the teaching features of the learning scenario to obtain the teacher interaction features. For example, this matching can be based on similarity metrics or matching algorithms such as cosine similarity and dynamic time warping.

[0122] In step S1023, the feature extraction of the teacher's knowledge scenario teaching features is performed to obtain a set of teacher's knowledge scenario teaching features corresponding to the historical interaction data, which specifically includes the following steps:

[0123] Obtain the number of knowledge scenario features in the knowledge scenario teaching feature set and the number of times the knowledge scenario features are learned and used;

[0124] The comprehensive learning degree of the knowledge scenario features is calculated based on the number of knowledge scenario features and the number of times the knowledge scenario features are learned and used;

[0125] If the comprehensive learning degree of the knowledge scenario feature meets the preset learning degree, the learning scenario teaching feature set corresponding to the knowledge scenario feature is obtained, and the teacher knowledge scenario teaching feature set is generated.

[0126] These feature sets can be used to evaluate teachers' teaching effectiveness, identify teaching patterns, and provide personalized training recommendations. This systematic approach extracts valuable features from teachers' knowledge scenarios and teaching characteristics, helping them better understand their teaching processes and outcomes, enabling targeted improvements and optimization.

[0127] In some embodiments, the above method of the present application may further include the steps of:

[0128] Teachers use the VR scene editor to obtain the created virtual teaching scene data; the virtual teaching scene data includes three-dimensional space coordinate data, teaching props layout data and dynamic interaction event configuration data;

[0129] Obtain the number of props, interaction node density, and spatial segmentation dimension, and calculate the scene complexity coefficient based on the number of props, interaction node density, and spatial segmentation dimension;

[0130] Dynamically couple the scene complexity coefficient with the training adjustment coefficient to generate the environmental fitness parameter α; the details are as follows:

[0131]

[0132] Among them, s_c is the scene complexity coefficient, t_c is the training adjustment coefficient, , is a learnable parameter, is the bias term, is the sigmoid function;

[0133] The teacher's interaction characteristics, teacher's knowledge scenario teaching characteristics, mind map characteristics and environmental adaptability parameters are input into the preset intelligent training model for training to obtain an AI large model, thereby achieving adaptive optimization of VR scene complexity and teaching training.

[0134] In some embodiments, the above method of the present application may further include the steps of:

[0135] Construct virtual-reality fusion training verification module:

[0136] Embed knowledge graph anchors in VR scenes, triggering multimodal explanations of related knowledge points when students interact with specific teaching props;

[0137] Collect students' attention distribution data through eye tracking and gesture recognition;

[0138] The real-time attention heat map is compared with the preset knowledge importance distribution map to generate scenario teaching effectiveness evaluation indicators and feed them back to the AI ​​big model. Through feature feedback and quantitative evaluation indicators, a closed-loop optimization system for teaching scenarios is constructed, which helps to dynamically adjust AI training strategies and improve model iteration efficiency.

[0139] Preferably, the scenario teaching effectiveness evaluation indicators include:

[0140] 1. Spatial cognitive efficiency index: calculated based on first fixation time and number of repeated visits;

[0141] 2. Operation logic matching: evaluated by the DTW distance between the actual interaction path and the preset teaching path;

[0142] 3. Knowledge conversion rate: calculated by combining the difference in pre- and post-test scores and the time spent in the scene.

[0143] Compared with the existing technology, the present application provides a method for training a large AI model for teaching in vocational schools, which obtains historical interaction data when a terminal device trains knowledge scenario teaching resources in a preset knowledge scenario teaching resource library; extracts features based on the historical interaction data to obtain teacher feature information corresponding to the historical interaction data; wherein the teacher feature information includes teacher interaction features and teacher knowledge scenario teaching features; obtains a training adjustment coefficient based on the teacher interaction features; generates a mind map related to the knowledge scenario teaching resources, and extracts mind map features from the mind map; inputs the teacher interaction features, teacher knowledge scenario teaching features, mind map features, and training adjustment coefficients into a preset intelligent training model for training to obtain an AI large model; Among them, the preset intelligent training model includes a feature data processing module, a teacher knowledge scenario teaching feature calculation module, a mind map feature calculation module, a teacher interaction feature calculation module, a comprehensive analysis module and a training result output module; the feature data processing module is used to perform feature extraction and data standardization processing on the teacher interaction feature, and obtain the first feature standardized data value of the teacher interaction feature and the second feature standardized data value of the teacher interaction feature; the teacher knowledge scenario teaching feature calculation module is used to perform feature extraction and data standardization processing on the teacher knowledge scenario teaching feature, and obtain the third feature standardized data value of the teacher knowledge scenario teaching feature; the mind map feature calculation module includes a sample representation component, a time representation component and a splicing component; The sample characterization component is used to extract node features, edge features and structural features, the time characterization component is used to extract time series features and time-dependent features from historical interaction data, and the splicing component is used to splice the node features, edge features and structural features extracted by the sample characterization component with the time series features and time-dependent features extracted by the time characterization component to form a feature vector, and use the feature vector as the merged data value; the teacher interaction feature calculation module is used to calculate based on the first feature standardized data value and the second feature standardized data value to obtain the comprehensive feature training value of the teacher interaction feature; the comprehensive analysis module is used to perform comprehensive processing on the third feature standardized data value, the comprehensive feature training value and the merged data value to obtain the optimal training data value ; The training result output module is used to obtain the preset AI big model training data information according to the optimal training data value, and train the AI ​​big model; the above method can not only accurately predict and better adapt to the differences of each teacher user and dynamically adapt to the changing data conditions, improve the matching degree between teaching courses and teachers' actual needs, better meet the personalized adaptation capabilities of teaching resources, and help improve the pertinence of AI big models to vocational school teaching scenarios, but also effectively optimize training efficiency, based on the dynamic adjustment mechanism of interactive matching degree, reduce invalid training data, improve model convergence speed, and improve resource utilization through comprehensive learning degree screening of knowledge scenario characteristics, give priority to the use of high-frequency teaching resources, and optimize resource allocation.

[0144] In the above embodiment, a vocational school teaching AI large model training system is provided. Correspondingly, this application also provides a vocational school teaching AI large model training system. Figure 2 As shown, the vocational school teaching AI large model training system provided in the embodiment of the present application can implement the above method. The vocational school teaching AI large model training system can be implemented by software, hardware, or a combination of software and hardware. For example, the vocational school teaching AI large model training system can include integrated or separate functional modules or units to perform the corresponding steps in the above methods, including:

[0145] The first acquisition unit 101 is used to acquire historical interaction data when the terminal device is training the knowledge scenario teaching resources in the preset knowledge scenario teaching resource library;

[0146] Extraction unit 102, configured to extract features from the historical interaction data to obtain teacher feature information corresponding to the historical interaction data; wherein the teacher feature information includes teacher interaction features and teacher knowledge scenario teaching features;

[0147] A second acquiring unit 103 is configured to acquire a training adjustment coefficient based on the teacher interaction characteristics;

[0148] A generating unit 104 is used to generate a mind map related to the knowledge scenario teaching resources and extract mind map features from the mind map;

[0149] The training unit 105 inputs the teacher interaction features, the teacher knowledge scenario teaching features, the mind map features and the training adjustment coefficient into a preset intelligent training model for training to obtain an AI large model;

[0150] Among them, the preset intelligent training model includes a feature data processing module, a teacher knowledge scenario teaching feature calculation module, a mind map feature calculation module, a teacher interaction feature calculation module, a comprehensive analysis module and a training result output module;

[0151] The feature data processing module is used to perform feature extraction and data standardization processing on the teacher interaction feature to obtain a first feature standardized data value of the teacher interaction feature and a second feature standardized data value of the teacher interaction feature;

[0152] The teacher knowledge scenario teaching feature calculation module is used to perform feature extraction and data standardization processing on the teacher knowledge scenario teaching feature to obtain the third feature standardized data value of the teacher knowledge scenario teaching feature;

[0153] The mind map feature calculation module includes a sample representation component, a time representation component, and a splicing component; the sample representation component is used to extract node features, edge features, and structural features; the time representation component is used to extract time series features and time-dependent features from historical interaction data; the splicing component is used to splice the node features, edge features, and structural features extracted by the sample representation component with the time series features and time-dependent features extracted by the time representation component to form a feature vector, and use the feature vector as the merged data value;

[0154] The teacher interaction feature calculation module is used to calculate the comprehensive feature training value of the teacher interaction feature based on the first feature standardized data value and the second feature standardized data value;

[0155] The comprehensive analysis module is used to comprehensively process the third feature standardized data value, the comprehensive feature training value and the merged data value to obtain the optimal training data value;

[0156] The training result output module is used to obtain the preset AI large model training data information based on the optimal training data value and train the AI ​​large model.

[0157] The system provided in the embodiment of the present application and the method provided in the embodiment of the present application are based on the same inventive concept and have the same beneficial effects as the methods adopted, operated or implemented by them.

[0158] An embodiment of the present application also provides an electronic device corresponding to the method provided in the above embodiment. The electronic device can be an electronic device used for a client, such as a mobile phone, a laptop computer, a tablet computer, a desktop computer, etc., to execute the above method.

[0159] Please refer to Figure 3 , which shows a schematic diagram of an electronic device provided by some embodiments of the present application. Figure 3 As shown, the electronic device 20 includes: a processor 200, a memory 201, a bus 202 and a communication interface 203, and the processor 200, the communication interface 203 and the memory 201 are connected via the bus 202; the memory 201 stores a computer program that can be run on the processor 200, and when the processor 200 runs the computer program, it executes the method provided in any of the aforementioned embodiments of the present application.

[0160] Memory 201 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage device. Communication between the system network element and at least one other network element is achieved through at least one communication interface 203 (which may be wired or wireless), and may utilize the Internet, a wide area network, a local area network, a metropolitan area network, or the like.

[0161] The bus 202 may be an ISA bus, a PCI bus, or an EISA bus. The bus may be divided into an address bus, a data bus, a control bus, etc. The memory 201 is used to store programs. The processor 200 executes the programs upon receiving execution instructions. The methods disclosed in any of the aforementioned embodiments of the present application may be applied to or implemented by the processor 200.

[0162] The processor 200 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by hardware integrated logic circuits in the processor 200 or by software instructions. The above processor 200 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in conjunction with the embodiments of this application can be directly implemented and executed by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium mature in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, etc. The storage medium is located in the memory 201 , and the processor 200 reads the information in the memory 201 and completes the steps of the above method in combination with its hardware.

[0163] The electronic device provided in the embodiments of the present application and the method provided in the embodiments of the present application are based on the same inventive concept and have the same beneficial effects as the methods adopted, operated or implemented by them.

[0164] The embodiments of the present application also provide a computer-readable storage medium corresponding to the method provided in the aforementioned embodiments, on which a computer program (i.e., a program product) is stored. When the computer program is run by a processor, it will execute the method provided in any of the aforementioned embodiments.

[0165] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical or magnetic storage media, which are not listed here one by one.

[0166] The computer-readable storage medium provided in the above-mentioned embodiments of the present application and the method provided in the embodiments of the present application are based on the same inventive concept and have the same beneficial effects as the method adopted, run or implemented by the application program stored therein.

[0167] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present application, and they should all be included in the scope of the claims and description of the present application.

Claims

1. A large-scale AI model training method for vocational school teaching, characterized in that: The following steps are involved: Obtain historical interaction data when the terminal device is training the knowledge scenario teaching resources in the preset knowledge scenario teaching resource library; Extracting features from historical interaction data to obtain teacher feature information corresponding to the historical interaction data; wherein the teacher feature information includes teacher interaction features and teacher knowledge scenario teaching features; Obtaining training adjustment coefficients based on teacher interaction characteristics; Generate a mind map related to knowledge scenario teaching resources and extract mind map features from the mind map; Input teacher interaction features, teacher knowledge scenario teaching features, mind map features, and training adjustment coefficients into the preset intelligent training model for training to obtain an AI large model; Among them, the preset intelligent training model includes a feature data processing module, a teacher knowledge scenario teaching feature calculation module, a mind map feature calculation module, a teacher interaction feature calculation module, a comprehensive analysis module and a training result output module; The feature data processing module is used to perform feature extraction and data standardization processing on the teacher interaction feature to obtain a first feature standardized data value of the teacher interaction feature and a second feature standardized data value of the teacher interaction feature; The teacher knowledge scenario teaching feature calculation module is used to perform feature extraction and data standardization processing on the teacher knowledge scenario teaching feature to obtain the third feature standardized data value of the teacher knowledge scenario teaching feature; The mind map feature calculation module includes a sample representation component, a time representation component, and a splicing component; the sample representation component is used to extract node features, edge features, and structural features; the time representation component is used to extract time series features and time-dependent features from historical interaction data; the splicing component is used to splice the node features, edge features, and structural features extracted by the sample representation component with the time series features and time-dependent features extracted by the time representation component to form a feature vector, and use the feature vector as the merged data value; The teacher interaction feature calculation module is used to calculate the comprehensive feature training value of the teacher interaction feature based on the first feature standardized data value and the second feature standardized data value; The comprehensive analysis module is used to comprehensively process the third feature standardized data value, the comprehensive feature training value and the merged data value to obtain the optimal training data value; The training result output module is used to obtain the preset AI large model training data information based on the optimal training data value, and train the AI ​​large model; The method of obtaining the training adjustment coefficient according to the teacher interaction characteristics specifically includes the following steps: Obtaining the interaction standard matching coefficient corresponding to the teacher interaction feature according to the teacher interaction feature; Obtain the comprehensive interaction matching degree corresponding to the teacher's interaction characteristics according to the interaction standard matching coefficient corresponding to the teacher's interaction characteristics; Compare the comprehensive interaction matching degree corresponding to the teacher's interaction feature with the interaction threshold. If the comprehensive interaction matching degree corresponding to the teacher's interaction feature is less than the interaction threshold, generate an interaction adjustment correction request corresponding to the teacher's interaction feature. The training adjustment coefficient is calculated based on the interactive adjustment correction request.

2. A vocational school teaching AI large model training method according to claim 1, characterized in that: The method of extracting features from historical interaction data to obtain teacher feature information corresponding to the historical interaction data specifically includes the following steps: Performing data classification on the historical interaction data to obtain an interaction information set corresponding to the historical interaction data; wherein the interaction information set includes teacher interaction information and teacher knowledge scenario teaching feature set; Extract features of teacher interaction information in the interaction information set to obtain teacher interaction features corresponding to historical interaction data; Feature extraction is performed on the teacher knowledge scenario teaching feature set in the interactive information set to obtain the teacher knowledge scenario teaching feature set corresponding to the historical interactive data.

3. The method for training a large AI model for teaching in vocational schools according to claim 1 is characterized in that: The step of calculating the training adjustment coefficient according to the interactive adjustment correction request specifically includes the following steps: Identifying the interaction adjustment and correction request and obtaining the terminal device interaction characteristics corresponding to the interaction adjustment and correction request; wherein the terminal device interaction characteristics include terminal device interaction habits and / or terminal device interaction time; The terminal interaction feature recognition degree corresponding to the interactive adjustment correction request is calculated based on the terminal device interaction feature corresponding to the interactive adjustment correction request; Comparing the terminal interaction feature recognition degree with the interaction adjustment interval, and generating a first adjustment instruction if the terminal interaction feature recognition degree is within a first preset adjustment interval; If the terminal interaction feature recognition degree is within a second preset adjustment range, generating a second adjustment instruction; If the terminal feature recognition degree exceeds the second preset adjustment range, a third adjustment instruction is generated; wherein the first adjustment instruction means no adjustment processing; the second adjustment instruction means partial adjustment processing; and the third adjustment instruction means full adjustment processing; The training adjustment coefficient is obtained by calculating the adjustment degrees corresponding to the first adjustment instruction, the second adjustment instruction, and the third adjustment instruction.

4. The method for training a large AI model for teaching in vocational schools according to claim 1, characterized in that: The comprehensive analysis module performs comprehensive processing on the third feature normalized data value, the comprehensive feature training value and the merged data value to obtain the optimal training data value, specifically comprising the following steps: Obtain the knowledge scenario teaching feature weight corresponding to the teacher's knowledge scenario teaching feature; The knowledge scenario teaching feature training value corresponding to the teacher's knowledge scenario teaching feature is calculated based on the knowledge scenario teaching feature weight and the third feature standardized data value; Obtain the adjustment coefficient weight corresponding to the teacher interaction characteristics; The teacher interaction feature training value corresponding to the teacher interaction feature is obtained according to the adjustment coefficient weight and the comprehensive feature training value; The optimal training data value is obtained by comprehensive calculation based on the knowledge scenario teaching feature training value, teacher interaction feature training value and merged data value.

5. The method for training a large AI model for teaching in vocational schools according to claim 2 is characterized in that: The feature extraction of the teacher interaction information in the interaction information set to obtain the teacher interaction features corresponding to the historical interaction data specifically includes the following steps: Data classification is performed on the interactive information set to obtain teacher interactive information and a knowledge scenario teaching feature set; wherein the knowledge scenario teaching feature set refers to teacher learning feature information corresponding to the knowledge scenario features under the teaching scenario; Converting the teacher learning features corresponding to the knowledge scenario features in the knowledge scenario teaching feature set into knowledge scenario teaching feature learning values; The teacher interaction information is matched with the learning values ​​of the teaching features of the learning scenario to obtain the teacher interaction features.

6. The method for training a large AI model for teaching in vocational schools according to claim 2 is characterized in that: The feature extraction of the teacher's knowledge scenario teaching features to obtain the teacher's knowledge scenario teaching feature set corresponding to the historical interaction data specifically includes the following steps: Obtain the number of knowledge scenario features in the knowledge scenario teaching feature set and the number of times the knowledge scenario features are learned and used; The comprehensive learning degree of the knowledge scenario features is calculated based on the number of knowledge scenario features and the number of times the knowledge scenario features are learned and used; If the comprehensive learning degree of the knowledge scenario feature meets the preset learning degree, the learning scenario teaching feature set corresponding to the knowledge scenario feature is obtained, and the teacher knowledge scenario teaching feature set is generated.

7. The method for training a large AI model for teaching in vocational schools according to claim 1, characterized in that: Also includes: Teachers use the VR scene editor to obtain the created virtual teaching scene data; the virtual teaching scene data includes three-dimensional space coordinate data, teaching props layout data and dynamic interaction event configuration data; Obtain the number of props, interaction node density, and spatial segmentation dimension, and calculate the scene complexity coefficient based on the number of props, interaction node density, and spatial segmentation dimension; Dynamically couple the scene complexity coefficient with the training adjustment coefficient to generate the environmental fitness parameter; The teacher interaction characteristics, teacher knowledge scenario teaching characteristics, mind map characteristics and environmental adaptability parameters are input into the preset intelligent training model for training to obtain the AI ​​large model.

8. The method for training a large AI model for teaching in vocational schools according to claim 6 is characterized in that: Also includes: Construct virtual-reality fusion training verification module: Embed knowledge graph anchors in VR scenes, triggering multimodal explanations of related knowledge points when students interact with specific teaching props; Collect students' attention distribution data through eye tracking and gesture recognition; Compare the real-time attention heat map with the preset knowledge importance distribution map to generate scenario teaching effectiveness evaluation indicators and feed them back to the AI ​​big model.

9. A large-scale AI model training system for vocational school teaching, characterized by: include: The first acquisition unit is used to acquire historical interaction data when the terminal device is training the knowledge scenario teaching resources in the preset knowledge scenario teaching resource library; An extraction unit is configured to extract features from the historical interaction data to obtain teacher feature information corresponding to the historical interaction data; wherein the teacher feature information includes teacher interaction features and teacher knowledge scenario teaching features; A second acquisition unit is used to obtain a training adjustment coefficient according to the teacher interaction characteristics; A generation unit, used to generate a mind map related to the knowledge scenario teaching resources and extract mind map features from the mind map; The training unit inputs the teacher interaction characteristics, teacher knowledge scenario teaching characteristics, mind map characteristics and training adjustment coefficients into the preset intelligent training model to train and obtain the AI ​​large model; Among them, the preset intelligent training model includes a feature data processing module, a teacher knowledge scenario teaching feature calculation module, a mind map feature calculation module, a teacher interaction feature calculation module, a comprehensive analysis module and a training result output module; The feature data processing module is used to perform feature extraction and data standardization processing on the teacher interaction feature to obtain a first feature standardized data value of the teacher interaction feature and a second feature standardized data value of the teacher interaction feature; The teacher knowledge scenario teaching feature calculation module is used to perform feature extraction and data standardization processing on the teacher knowledge scenario teaching feature to obtain the third feature standardized data value of the teacher knowledge scenario teaching feature; The mind map feature calculation module includes a sample representation component, a time representation component, and a splicing component; the sample representation component is used to extract node features, edge features, and structural features; the time representation component is used to extract time series features and time-dependent features from historical interaction data; the splicing component is used to splice the node features, edge features, and structural features extracted by the sample representation component with the time series features and time-dependent features extracted by the time representation component to form a feature vector, and use the feature vector as the merged data value; The teacher interaction feature calculation module is used to calculate the comprehensive feature training value of the teacher interaction feature based on the first feature standardized data value and the second feature standardized data value; The comprehensive analysis module is used to comprehensively process the third feature standardized data value, the comprehensive feature training value and the merged data value to obtain the optimal training data value; The training result output module is used to obtain the preset AI large model training data information based on the optimal training data value, and train the AI ​​large model; The second acquiring unit is further configured to: Obtaining the interaction standard matching coefficient corresponding to the teacher interaction feature according to the teacher interaction feature; Obtain the comprehensive interaction matching degree corresponding to the teacher's interaction characteristics according to the interaction standard matching coefficient corresponding to the teacher's interaction characteristics; Compare the comprehensive interaction matching degree corresponding to the teacher's interaction feature with the interaction threshold. If the comprehensive interaction matching degree corresponding to the teacher's interaction feature is less than the interaction threshold, generate an interaction adjustment correction request corresponding to the teacher's interaction feature. The training adjustment coefficient is calculated based on the interactive adjustment correction request.

Images