A teaching method for ideological and political practice tasks in a virtual-real interaction scene

Abstract

The present application relates to the technical field of education and teaching system, and particularly relates to a kind of ideological and political practice task teaching method for virtual and real interactive scene, it obtains the multi-modal behavior data of student in ideological and political practice task virtual interactive scene and carries out feature extraction, obtains the learning state feature of student, then according to learning state feature, obtain the emotional value grade of student, wherein emotional value grade is used to indicate the emotional resonance degree and value identification degree of student in practice learning process, finally according to emotional value grade, obtain the teaching effect score of student.The present application constructs a full range, three-dimensional student state perception system by collecting multi-modal data covering physiological response, behavior trajectory and language expression, realizes the objective quantification of emotional value in the practice of ideological and political education virtual scene, can realize scientific, accurate teaching feedback, solves the problem that current ideological and political practice task teaching method for virtual and real interactive scene cannot automatically evaluate learning effect.

Description

Technical Field

[0001] This invention relates to the field of educational and teaching systems, specifically to a teaching method for ideological and political practice tasks in a virtual-real interactive scenario. Background Technology

[0002] With the rapid development of virtual reality (VR), augmented reality (AR), and mixed reality (MR) technologies, virtual-real interactive scenarios have gradually become an important carrier for ideological and political education practice. Compared with traditional classroom teaching or field visits, this immersive and embodied interactive environment can break the limitations of time and space, highly reproduce complex real-world situations, and provide students with an "experiential" learning experience.

[0003] However, there is a fundamental difference in educational goals between ideological and political education and natural science disciplines. Natural sciences focus on the cognition of objective facts and the achievement of skill indicators, while the core of ideological and political education lies in guiding worldviews, outlooks on life, and values. Its emphasis is on stimulating and internalizing emotional values. It generally lacks explicit quantitative indicators to evaluate students' learning outcomes, making it impossible to achieve automated scoring of teaching effectiveness. Especially in ideological and political practice tasks in interactive virtual and real-world scenarios, relying on subjective human observation or after-class summaries to assess students' learning outcomes is extremely inefficient.

[0004] Therefore, a new teaching method is needed to address the problem that current teaching methods for ideological and political practice tasks in virtual-real interactive scenarios cannot automatically assess learning outcomes. Summary of the Invention

[0005] The purpose of this invention is to provide a teaching method for ideological and political practice tasks in virtual-real interactive scenarios, and to solve the following technical problems:

[0006] Current teaching methods for ideological and political practice tasks in virtual-real interactive scenarios cannot automatically evaluate learning outcomes.

[0007] The objective of this invention can be achieved through the following technical solutions:

[0008] A teaching method for ideological and political practice tasks in virtual-real interactive scenarios includes the following steps:

[0009] Acquire multimodal behavioral data of students in virtual interactive scenarios of ideological and political practice tasks, including physiological data, behavioral data and expressive data;

[0010] Feature extraction is performed on multimodal behavioral data to obtain the learning state characteristics of students;

[0011] Based on the characteristics of the learning state, the students' emotional value level is obtained, where the emotional value level is used to represent the students' emotional resonance and value recognition during the practical learning process;

[0012] Based on the emotional value level, students' learning effectiveness is rated.

[0013] As a further aspect of this invention: learning state characteristics include emotional resonance characteristics and value identification characteristics; feature extraction is performed on multimodal behavioral data to obtain students' learning state characteristics, including:

[0014] Feature extraction is performed on physiological and behavioral data to obtain emotional resonance features;

[0015] Feature extraction is performed on behavioral and expressive data to obtain value recognition features.

[0016] As a further aspect of the present invention: physiological data includes heart rate data, eye movement data, and facial expression data; behavioral data includes decision-making behavior data; feature extraction is performed on the physiological and behavioral data to obtain emotional resonance features, including:

[0017] Physiological intensity characteristics are obtained based on the fluctuations in heart rate data;

[0018] Attention focus characteristics are obtained based on the duration of gaze fixation in eye-tracking data;

[0019] Based on the proportion of frames containing different facial expressions in the facial expression data, facial expression matching features are obtained;

[0020] Emotional response characteristics are obtained based on the hesitation time in decision-making behavior data;

[0021] By summarizing physiological intensity characteristics, attention focus characteristics, facial expression matching characteristics, and emotional response characteristics, we can obtain emotional resonance characteristics.

[0022] As a further aspect of the present invention: behavioral data includes decision content data; feature extraction is performed on the behavioral data and expression data to obtain value recognition features, including:

[0023] Based on the distribution of decision content in the decision content data, the statistical characteristics of decision-making are obtained;

[0024] Based on the frequency of keywords indicating agreement in the expression data, the degree of agreement characteristics are obtained;

[0025] By summarizing the statistical characteristics of decision-making and the characteristics of the degree of recognition, we can obtain the characteristics of value recognition.

[0026] As a further aspect of the present invention: based on the characteristics of the learning state, the student's emotional value level is obtained, including:

[0027] Input data is obtained based on the characteristics of emotional resonance and value identification;

[0028] Input data is fed into a preset artificial intelligence analysis model to obtain the emotional value level output by the artificial intelligence analysis model. The emotional value level includes emotional resonance level and value recognition level.

[0029] As a further aspect of the present invention: the artificial intelligence model includes an input layer and a primary neural network layer connected thereto. The primary neural network layer is simultaneously connected to a first neural network branch and a second neural network branch. The first neural network branch is used to output the emotional resonance level, and the second neural network branch is used to output the value recognition level.

[0030] As a further aspect of the present invention, it also includes training a pre-defined artificial intelligence model, which specifically includes:

[0031] Obtain labeled and unlabeled data, where labeled data represents learning state features with labeled sentiment value levels, and unlabeled data represents learning state features without labeled sentiment value levels;

[0032] Based on the labeled data, the initial cluster centers for each emotional value level are obtained;

[0033] Based on the initial cluster centers, labeled and unlabeled data are clustered to obtain multiple clusters that correspond one-to-one with multiple sentiment value levels;

[0034] The sentiment value level corresponding to the cluster where the unlabeled data is located is used as its corresponding label to complete the labeling of the unlabeled data and obtain the training data.

[0035] A pre-set artificial intelligence model is trained based on training data.

[0036] A teaching system for ideological and political practice tasks in virtual-real interactive scenarios, comprising:

[0037] The multimodal monitoring module is used to acquire multimodal behavioral data of students in virtual interactive scenarios of ideological and political practice tasks. The multimodal behavioral data includes physiological data, behavioral data and expressive data.

[0038] The feature extraction module is used to extract features from multimodal behavioral data to obtain the learning status features of students;

[0039] The emotional value analysis module is used to obtain the emotional value level of students based on the characteristics of their learning status. The emotional value level is used to represent the degree of emotional resonance and value recognition of students in the process of practical learning.

[0040] The learning assessment module is used to obtain a score of students' learning effectiveness based on their affective value level.

[0041] An electronic device, comprising:

[0042] Memory and processor;

[0043] The memory is used to store the program, and the processor is used to execute any of the steps in the teaching method for ideological and political practice tasks in virtual and real interactive scenarios when the program is executed.

[0044] A computer-readable storage medium for storing a computer-readable program or instructions, which, when executed by a processor, can implement any of the steps in the teaching method for ideological and political practice tasks in virtual-real interactive scenarios.

[0045] The beneficial effects of this invention are:

[0046] This invention provides a teaching method for ideological and political practice tasks in virtual-real interactive scenarios. It acquires multimodal behavioral data of students in these scenarios, including physiological, behavioral, and expressive data. Feature extraction is then performed on this data to obtain students' learning state characteristics. Based on these characteristics, an emotional value level is calculated, representing the student's emotional resonance and value identification during the learning process. Finally, a teaching effectiveness score is derived based on the emotional value level. This invention constructs a comprehensive and three-dimensional student state perception system by collecting multimodal data encompassing physiological reactions, behavioral trajectories, and verbal expressions. This enables the objective quantification of emotional value in virtual ideological and political education practices, providing scientific and precise teaching feedback and solving the problem that current teaching methods for ideological and political practice tasks in virtual-real interactive scenarios cannot automatically assess learning effectiveness. Attached Figure Description

[0047] The invention will now be further described with reference to the accompanying drawings.

[0048] Figure 1 This is a flowchart of the teaching method for ideological and political practice tasks in virtual-real interactive scenarios according to the present invention;

[0049] Figure 2 This is a system architecture diagram of the ideological and political practice task teaching system for virtual and real interactive scenarios according to the present invention. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] Please see Figure 1As shown, this invention is a teaching method for ideological and political practice tasks in virtual-real interactive scenarios, comprising the following steps:

[0052] S101. Obtain multimodal behavioral data of students in virtual interactive scenarios of ideological and political practice tasks, including physiological data, behavioral data and expressive data.

[0053] S102. Extract features from multimodal behavioral data to obtain students' learning status features;

[0054] S103. Based on the characteristics of the learning state, the emotional value level of the students is obtained, where the emotional value level is used to represent the students' emotional resonance and value recognition in the process of practical learning.

[0055] S104. Based on the emotional value level, obtain the students' teaching effectiveness score.

[0056] In the above process, multimodal behavioral data refers to any data that can be collected in virtual interactive teaching scenarios, including students' physiological data, behavioral data, and expressive data. Specifically, physiological data refers to data that can reflect students' physiological emotions, such as heart rate, skin conductance response, eye movement trajectory, and micro-expressions. Behavioral data refers to data that represents students' interactive behaviors, such as decision options, operation trajectories, and action content. Expressive data refers to data with semantic content, such as students' spoken dialogues, input, or text content selected in the interactive system.

[0057] It is understandable that in virtual interaction scenarios, students need to wear virtual reality devices, so collecting the aforementioned physiological and behavioral data is a natural and easy method to implement without significantly impacting student learning. Those skilled in the art can also conceive of how to achieve this. Furthermore, obtaining this data would naturally require prior notification to students, thus avoiding privacy or data security issues.

[0058] Emotional value represents the degree to which students internalize ideological and political education. In this embodiment, it is judged from two dimensions: emotional resonance and value recognition. It is understandable that emotional value is difficult to quantify with specific numerical values ​​in practice. Therefore, this embodiment uses a relatively vague labeling of emotional value levels (such as high, medium, and low) to facilitate automated scoring. The specific implementation method will be described in detail later.

[0059] Specifically, in one embodiment, the learning state characteristics are also divided into two dimensions: emotional resonance characteristics and value identification characteristics. Based on this, step S102 above, which extracts features from the multimodal behavioral data to obtain the student's learning state characteristics, specifically includes:

[0060] Feature extraction is performed on physiological and behavioral data to obtain emotional resonance features;

[0061] Feature extraction is performed on behavioral and expressive data to obtain value recognition features.

[0062] This embodiment captures the intensity of students' emotional engagement with virtual scenarios from the levels of physiological instincts and subconscious behavior through physiological and behavioral data. This avoids the one-sidedness of judging solely based on surface behavior, obtaining emotional resonance characteristics. By combining behavioral and expressive data, it directly maps students' cognitive understanding and active acceptance of ideological and political values, reflecting value identification characteristics. This multi-dimensional feature extraction strategy not only clarifies the internal logical chain of emotional value but also provides highly discriminative data support for subsequent determination of emotional value levels.

[0063] It is conceivable that different types of data acquired in practice will lead to different specific methods of feature extraction, which can be flexibly designed according to actual needs by those skilled in the art. This invention provides a more concrete example to explain the above feature extraction process:

[0064] In one embodiment, physiological data includes heart rate data, eye movement data (eye movement trajectory), and facial expression data; behavioral data includes decision-making behavior data (specifically, the hesitation time and number of changes when making a decision). Based on this, the step is to extract features from the physiological and behavioral data to obtain emotional resonance features, specifically including:

[0065] Physiological intensity characteristics are obtained based on the fluctuations in heart rate data;

[0066] Attention focus characteristics are obtained based on the duration of gaze fixation in eye-tracking data;

[0067] Based on the proportion of frames containing different facial expressions in the facial expression data, facial expression matching features are obtained;

[0068] Emotional response characteristics are obtained based on the hesitation time in decision-making behavior data;

[0069] By summarizing physiological intensity characteristics, attention focus characteristics, facial expression matching characteristics, and emotional response characteristics, we can obtain emotional resonance characteristics.

[0070] Among these features, heart rate fluctuations can be statistically analyzed using any existing dimension such as heart rate variability rate and peak value, representing changes in students' emotions; eye-tracking data, such as gaze duration, can be represented by the time the eyes linger at a specific location in the virtual teaching scene, reflecting students' level of attention; facial expression data, such as the percentage of frames for different expressions (e.g., smile, surprise), reflects students' level of emotional resonance; and decision-making behavior data, such as hesitation time, represents students' emotional response speed, indirectly indicating their level of emotional engagement. After obtaining these features, encoding them into specific vectors yields the emotional resonance features.

[0071] Furthermore, in one embodiment, the behavioral data includes decision content data (specifically referring to the specific options, content, etc. of the decision). Based on this, the steps include: extracting features from the behavioral data and expression data to obtain value recognition features, specifically including:

[0072] Based on the distribution of decision content in the decision content data, the statistical characteristics of decision-making are obtained;

[0073] Based on the frequency of keywords indicating agreement in the expression data, the degree of agreement characteristics are obtained;

[0074] By summarizing the statistical characteristics of decision-making and the characteristics of the degree of recognition, we can obtain the characteristics of value recognition.

[0075] The distribution of decision content in the decision content data intuitively reflects the accuracy of value choices. Keywords indicating agreement in the expression data refer to any words that can express agreement, such as "agree," "can," and "yes." Statistical analysis of the frequency of these words can intuitively reveal the degree of value agreement. In practice, these words can be pre-defined, automatically identified through word embedding techniques, or statistically analyzed using more advanced semantic recognition models combined with context. All of these are existing technologies accessible to those skilled in the art.

[0076] The above embodiments provide a clear and feasible technical path. In the extraction of emotional resonance features, feature vectors are constructed from four dimensions: instinctive reaction, attention allocation, emotional expression, and response speed. This enables a refined quantification of students' emotional state. In the extraction of value recognition features, a dual verification mechanism of decision content and keyword frequency is cleverly utilized. This examines not only the students' behavioral outcomes in virtual choices but also their language expression in reflection, forming a closed-loop verification that integrates knowledge and action from behavior to cognition.

[0077] Understandably, after obtaining the above learning status characteristics, any method can be used to assess the level of affective value. The simplest method is to first pre-set standards for the learning status characteristics, and then compare the student's learning status characteristics with the pre-set standards to obtain the level of affective value. Taking value identification characteristics as an example, if the student's decision-making statistics and identification level characteristics both meet the standards, they can be judged as "deep identification" level; if only one characteristic meets the standard, they can be judged as "initial identification"; and if no characteristic meets the standard, they can be judged as "low identification".

[0078] Furthermore, the present invention provides a more preferred method for determining the level of emotional value: In a preferred embodiment, the above step S103, obtaining the student's emotional value level based on the learning state characteristics, specifically includes;

[0079] Input data is obtained based on the characteristics of emotional resonance and value identification;

[0080] Input data is fed into a preset artificial intelligence analysis model to obtain the emotional value level output by the artificial intelligence analysis model. The emotional value level includes emotional resonance level and value recognition level.

[0081] This embodiment employs an artificial intelligence model that can automatically learn the complex nonlinear relationship between emotional resonance features and value identification features, avoiding the subjectivity and limitations of manually setting fixed thresholds. This allows for a more accurate capture of students' subtle psychological changes and value orientations in virtual scenarios. Furthermore, this embodiment uses both types of features as input data simultaneously, constructing a more comprehensive evaluation system. Emotional resonance features reveal the depth of students' emotional engagement with ideological and political situations, while value identification features reflect their level of rational understanding of ideological and political concepts. The fusion of these two inputs enables the model to not only determine whether students are moved, but also whether they identify with the concepts because of that emotion, achieving a logically progressive evaluation from emotional arousal to value internalization. In addition, this embodiment designs the output data as qualitative levels such as "emotional resonance level" and "value identification level," rather than specific numerical scores, cleverly avoiding the problem of the difficulty in quantifying emotional value and facilitating the training and implementation of the artificial intelligence model.

[0082] It is understood that the aforementioned artificial intelligence model can employ any existing data-driven model, such as a neural network-based classifier or any other existing model. This invention provides a more preferred implementation:

[0083] In one embodiment, the artificial intelligence model includes an input layer and a primary neural network layer connected thereto. The primary neural network layer is simultaneously connected to a first neural network branch and a second neural network branch. The first neural network branch is used to output the emotional resonance level, and the second neural network branch is used to output the value recognition level.

[0084] The dual-branch neural network structure designed in this embodiment, while inheriting the powerful fitting capabilities of data-driven models, further optimizes the precise decoupling and evaluation of the dual objectives of ideological and political education. This structure uses a shared primary neural network layer to jointly extract features from the input "emotional resonance features" and "value identification features," fully exploring the potential deep correlation between the two types of features (such as how high emotional resonance promotes value identification). Subsequently, independent branch networks output "emotional resonance level" and "value identification level" respectively, achieving independent quantification and parallel analysis of the core elements of ideological and political education. This design effectively avoids the mutual interference that may occur when single-task models handle multi-dimensional complex objectives, improving the model's training efficiency and prediction accuracy.

[0085] Depending on the specific needs in practice, the neural network branch can adopt any existing model structure. One of the simplest ways is to connect multiple fully connected layers in series to obtain a multi-level simple feedforward neural network, which can then be used to obtain the first / second neural network branch.

[0086] Furthermore, the use of artificial intelligence models requires training. However, for the application scenario of this invention, acquiring a large amount of sample data and labeling this sample data with sentiment value levels is by no means easy. Therefore, this invention also provides a preferred method to improve the feasibility of the artificial intelligence model in this embodiment. Specifically, the teaching method for ideological and political practice tasks in virtual-real interaction scenarios in this embodiment further includes the step of training a preset artificial intelligence model, which specifically includes:

[0087] Obtain labeled and unlabeled data, where labeled data represents learning state features with labeled sentiment value levels, and unlabeled data represents learning state features without labeled sentiment value levels;

[0088] Based on the labeled data, the initial cluster centers for each emotional value level are obtained;

[0089] Based on the initial cluster centers, labeled and unlabeled data are clustered to obtain multiple clusters that correspond one-to-one with multiple sentiment value levels;

[0090] The sentiment value level corresponding to the cluster where the unlabeled data is located is used as its corresponding label to complete the labeling of the unlabeled data and obtain the training data.

[0091] A pre-set artificial intelligence model is trained based on training data.

[0092] Specifically, the labeled data mentioned above refers to data pre-labeled with emotional resonance levels (e.g., high, medium, and low) and value identification levels (e.g., deep identification, moderate identification, and initial identification). Unlabeled data refers to learning state features obtained through pilot projects or randomly generated without pre-labeled emotional value levels. The purpose of this embodiment is to automatically classify the levels of unlabeled data using only a small amount of labeled data and an unsupervised clustering algorithm. During the clustering process, the centers of labeled data at the same level can be selected as initial cluster centers. For example, the mean of all labeled data with high emotional resonance levels can be selected as the initial cluster center. Then, based on these cluster centers, a clustering algorithm based on a specified number of clusters (e.g., K-means) is used to cluster all unlabeled data similar to those with high emotional resonance levels into the same cluster. At this point, all data in that cluster can be considered as having high emotional resonance levels, thus achieving "pseudo-labeling" of the data and obtaining a massive amount of sample data suitable for training artificial intelligence models at minimal cost.

[0093] It is worth emphasizing that the above training process is made possible precisely because the present invention uses an evaluation method of "emotional value level" rather than "emotional value score". This is because unlabeled data in the same cluster only need to be judged in terms of level, without the need to calculate specific values.

[0094] This embodiment proposes an innovative semi-supervised clustering training method, significantly reducing the application threshold and implementation cost of artificial intelligence models. This method cleverly utilizes a small amount of manually labeled "seed data" to determine the initial cluster centers for sentiment value levels. Then, a clustering algorithm automatically classifies the feature vectors of massive amounts of unlabeled data, thereby achieving batch pseudo-labeling of unlabeled data. This process not only greatly reduces reliance on expensive expert annotation but also ensures the quality of pseudo-labels through cluster consistency constraints, enabling the model to converge quickly with limited annotation resources. As more student data accumulates, the model can continuously optimize its evaluation accuracy through continuous clustering updates. This embodiment perfectly aligns with the realities of sensitive data and difficult sample acquisition in ideological and political education, providing a feasible technical path for the large-scale implementation of sentiment value assessment technology.

[0095] Combination Figure 2 The present invention also provides a teaching system for ideological and political practice tasks in virtual-real interactive scenarios, comprising:

[0096] The multimodal monitoring module 210 is used to acquire multimodal behavioral data of students in virtual interactive scenarios of ideological and political practice tasks, including physiological data, behavioral data and expression data.

[0097] Feature extraction module 220 is used to extract features from multimodal behavioral data to obtain the learning state features of students;

[0098] The emotional value analysis module 230 is used to obtain the emotional value level of students based on the characteristics of their learning status. The emotional value level is used to represent the degree of emotional resonance and value recognition of students in the process of practical learning.

[0099] Learning assessment module 240 is used to obtain a score of students' learning effectiveness based on their affective value level.

[0100] The present invention also provides an electronic device, comprising:

[0101] Memory and processor;

[0102] The memory is used to store the program, and the processor is used to execute any of the steps in the teaching method for ideological and political practice tasks in virtual and real interactive scenarios when the program is executed.

[0103] The present invention also provides a computer-readable storage medium for storing a computer-readable program or instruction, wherein when the program or instruction is executed by a processor, it can implement any of the steps in the above-mentioned teaching method for ideological and political practice tasks in virtual-real interactive scenarios.

[0104] This invention provides a teaching method for ideological and political practice tasks in virtual-real interactive scenarios. It acquires multimodal behavioral data of students in these scenarios, including physiological, behavioral, and expressive data. Feature extraction is then performed on this data to obtain students' learning state characteristics. Based on these characteristics, an emotional value level is calculated, representing the student's emotional resonance and value identification during the learning process. Finally, a teaching effectiveness score is derived based on the emotional value level. This invention constructs a comprehensive and three-dimensional student state perception system by collecting multimodal data encompassing physiological reactions, behavioral trajectories, and verbal expressions. This enables the objective quantification of emotional value in virtual ideological and political education practices, providing scientific and precise teaching feedback and solving the problem that current teaching methods for ideological and political practice tasks in virtual-real interactive scenarios cannot automatically assess learning effectiveness.

[0105] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A teaching method for ideological and political practice tasks in virtual-real interactive scenarios, characterized in that, Includes the following steps: Acquire multimodal behavioral data of students in virtual interactive scenarios of ideological and political practice tasks. The multimodal behavioral data includes physiological data, behavioral data and expression data. Physiological data includes heart rate data, eye movement data and facial expression data. Behavioral data includes decision-making behavior data. Feature extraction is performed on multimodal behavioral data to obtain students' learning state characteristics, which include emotional resonance characteristics and value identification characteristics. Based on the characteristics of the learning state, the students' emotional value level is obtained, where the emotional value level is used to represent the students' emotional resonance and value recognition during the practical learning process; Based on the emotional value level, students' teaching effectiveness is rated. Among these, feature extraction is performed on multimodal behavioral data to obtain students' learning state features, including: Physiological intensity characteristics are obtained based on the fluctuations in heart rate data; Attention focus characteristics are obtained based on the duration of gaze fixation in eye-tracking data; Based on the proportion of frames containing different facial expressions in the facial expression data, facial expression matching features are obtained; Emotional response characteristics are obtained based on the hesitation time in decision-making behavior data; By summarizing physiological intensity characteristics, attention focus characteristics, facial expression matching characteristics, and emotional response characteristics, emotional resonance characteristics are obtained. Based on the distribution of decision content in the decision content data, the statistical characteristics of decision-making are obtained; Based on the frequency of keywords indicating agreement in the expression data, the degree of agreement characteristics are obtained; By summarizing the statistical characteristics of decision-making and the characteristics of the degree of recognition, we can obtain the characteristics of value recognition. Among them, based on the characteristics of the learning state, the students' emotional value level is obtained, including: Input data is obtained based on the characteristics of emotional resonance and value identification; Input data is fed into a preset artificial intelligence analysis model to obtain the emotional value level output by the artificial intelligence analysis model. The emotional value level includes emotional resonance level and value recognition level. The artificial intelligence model includes an input layer and a primary neural network layer connected to it. The primary neural network layer is followed by a first neural network branch and a second neural network branch. The first neural network branch is used to output the level of emotional resonance, and the second neural network branch is used to output the level of value recognition.

2. The teaching method for ideological and political practice tasks in virtual-real interactive scenarios according to claim 1, characterized in that, It also includes training a pre-defined artificial intelligence model, which specifically includes: Obtain labeled and unlabeled data, where labeled data represents learning state features with labeled sentiment value levels, and unlabeled data represents learning state features without labeled sentiment value levels; Based on the labeled data, the initial cluster centers for each emotional value level are obtained; Based on the initial cluster centers, labeled and unlabeled data are clustered to obtain multiple clusters that correspond one-to-one with multiple sentiment value levels; The sentiment value level corresponding to the cluster where the unlabeled data is located is used as its corresponding label to complete the labeling of the unlabeled data and obtain the training data. A pre-set artificial intelligence model is trained based on training data.

3. A teaching system for ideological and political practice tasks in virtual-real interactive scenarios, characterized in that, include: The multimodal monitoring module is used to acquire multimodal behavioral data of students in virtual interactive scenarios of ideological and political practice tasks. The multimodal behavioral data includes physiological data, behavioral data and expression data. Physiological data includes heart rate data, eye movement data and facial expression data, and behavioral data includes decision-making behavior data. The feature extraction module is used to extract features from multimodal behavioral data to obtain students' learning status features, which include emotional resonance features and value identification features. The emotional value analysis module is used to obtain the emotional value level of students based on the characteristics of their learning status. The emotional value level is used to represent the degree of emotional resonance and value recognition of students in the process of practical learning. The learning assessment module is used to obtain a score of students' learning effectiveness based on their affective value level. Among these, feature extraction is performed on multimodal behavioral data to obtain students' learning state features, including: Physiological intensity characteristics are obtained based on the fluctuations in heart rate data; Attention focus characteristics are obtained based on the duration of gaze fixation in eye-tracking data; Based on the proportion of frames containing different facial expressions in the facial expression data, facial expression matching features are obtained; Emotional response characteristics are obtained based on the hesitation time in decision-making behavior data; By summarizing physiological intensity characteristics, attention focus characteristics, facial expression matching characteristics, and emotional response characteristics, emotional resonance characteristics are obtained. Based on the distribution of decision content in the decision content data, the statistical characteristics of decision-making are obtained; Based on the frequency of keywords indicating agreement in the expression data, the degree of agreement characteristics are obtained; By summarizing the statistical characteristics of decision-making and the characteristics of the degree of recognition, we can obtain the characteristics of value recognition. Among them, based on the characteristics of the learning state, the students' emotional value level is obtained, including: Input data is obtained based on the characteristics of emotional resonance and value identification; Input data is fed into a preset artificial intelligence analysis model to obtain the emotional value level output by the artificial intelligence analysis model. The emotional value level includes emotional resonance level and value recognition level. The artificial intelligence model includes an input layer and a primary neural network layer connected to it. The primary neural network layer is followed by a first neural network branch and a second neural network branch. The first neural network branch is used to output the level of emotional resonance, and the second neural network branch is used to output the level of value recognition.

4. An electronic device, characterized in that, include: Memory and processor; The memory is used to store the program, and the processor is used to execute the steps in the teaching method for ideological and political practice tasks in virtual and real interactive scenarios as described in any one of claims 1-2 when executing the program.

5. A computer-readable storage medium, characterized in that, Used to store computer-readable programs or instructions, which, when executed by a processor, are capable of implementing the steps in the teaching method for ideological and political practice tasks in a virtual-real interactive scenario as described in any one of claims 1-2.

Images