Art education learning path generation method based on large-screen equipment and terminal

Abstract

The invention discloses a method and terminal for generating an art education learning path based on large-screen equipment, and the method comprises the steps: collecting static student data and dynamic interaction data of students in the large-screen equipment, and constructing art student portraits for the corresponding students through employing the static student data and the dynamic interaction data; and matching corresponding art learning contents in an art education resource library according to the artistic student portrait, and generating an art education learning path according to the artistic student portrait and the art learning contents. Therefore, the interactive data between the students and the large-screen equipment and the data of the students can be fully utilized to construct the artistic student portrait, and the learning path is generated in combination with the artistic student portrait and the generated artistic learning content, so that the individuation and high flexibility of the learning path are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of art education data processing, and in particular to a method and a terminal for generating an art education learning path based on a large-screen device. Background Art

[0002] At present, there are some intelligent learning systems in the field of art education, but the existing systems often lack personalization and interactivity, cannot fully utilize the advantages of large-screen devices, and have a poor interactive experience. It is also difficult to provide students with truly customized learning paths in intelligent learning systems, and it is difficult to adjust the learning paths in real time, which lacks flexibility. Summary of the invention

[0003] The technical problem to be solved by the present invention is: to provide a method and terminal for generating an art education learning path based on a large-screen device, which can generate a personalized and flexible art education learning path according to the learning information of students through the large-screen device.

[0004] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0005] A method for generating an art education learning path based on a large-screen device comprises the following steps:

[0006] Collecting art learning data of students on large-screen devices, the art learning data including static student data and dynamic interaction data, and using the art learning data to construct an art student portrait for the corresponding student;

[0007] Matching corresponding art learning content in an art education resource library according to the art student portrait;

[0008] An art education learning path is generated according to the art student portrait and the art learning content.

[0009] In order to solve the above technical problems, another technical solution adopted by the present invention is:

[0010] A terminal for generating an art education learning path based on a large-screen device comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, each step of the above-mentioned method for generating an art education learning path based on a large-screen device is implemented.

[0011] The beneficial effects of the present invention are: collecting static student data and dynamic interaction data of students in large-screen devices, using the static student data and dynamic interaction data to construct art student portraits for corresponding students; matching corresponding art learning content in an art education resource library according to the art student portraits, and generating art education learning paths according to the art student portraits and art learning content. Therefore, the interactive data between students and large-screen devices and the students' own data can be fully utilized to construct art student portraits, and learning paths can be generated by combining art student portraits and generated art learning content, ensuring the personalization and high flexibility of learning paths. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 A flowchart of a method for generating an art education learning path based on a large-screen device according to an embodiment of the present invention;

[0013] Figure 2 A schematic diagram of a terminal for generating an art education learning path based on a large-screen device according to an embodiment of the present invention;

[0014] Figure 3 A module diagram of a method for generating an art education learning path based on a large-screen device according to an embodiment of the present invention;

[0015] Figure 4 This is a main flow chart of a method for generating an art education learning path based on a large-screen device according to an embodiment of the present invention;

[0016] Description of labels:

[0017] 1. A terminal for generating art education learning paths based on large-screen devices; 2. Memory; 3. Processor. DETAILED DESCRIPTION

[0018] In order to explain the technical content, achieved objectives and effects of the present invention in detail, the following is an explanation in combination with the implementation modes and the accompanying drawings.

[0019] Please refer to Figure 1 The embodiment of the present invention provides a method for generating an art education learning path based on a large-screen device, comprising the steps of:

[0020] Collecting art learning data of students on large-screen devices, the art learning data including static student data and dynamic interaction data, and using the art learning data to construct an art student portrait for the corresponding student;

[0021] Matching corresponding art learning content in an art education resource library according to the art student portrait;

[0022] An art education learning path is generated according to the art student portrait and the art learning content.

[0023] From the above description, it can be seen that the beneficial effects of the present invention are: collecting static student data and dynamic interaction data of students in large-screen devices, using static student data and dynamic interaction data to build art student portraits for corresponding students; matching corresponding art learning content in the art education resource library according to the art student portraits, and generating art education learning paths according to the art student portraits and art learning content. Therefore, it is possible to fully utilize the interactive data between students and large-screen devices and the students' own data to build art student portraits, and generate learning paths in combination with art student portraits and generated art learning content, ensuring the personalization and high flexibility of learning paths.

[0024] Furthermore, using the art learning data to construct an art student portrait for the corresponding student includes:

[0025] The static student data includes historical learning data, and the dynamic interaction data includes facial expression data, body movement data and interaction instruction data;

[0026] Use a deep learning model to extract first feature data of the facial expression data and the body movement data, use a natural language processing model to analyze the interaction instruction data to obtain second feature data, and use time series analysis to process the historical learning data to obtain third feature data;

[0027] Performing weighted fusion on the first feature data, the second feature data and the third feature data to obtain a fused feature;

[0028] Performing cluster analysis on the fusion features to obtain the learning type of the student corresponding to the art learning data, the learning type including learning preference type, learning style type and student level type, and generating a first student portrait according to the learning type;

[0029] According to the art learning data of students acquired in real time, the first student portrait is updated in real time using an online learning algorithm to obtain a real-time art student portrait.

[0030] From the above description, it can be seen that by combining the feature extraction results of dynamic and static data for cluster analysis to provide a more comprehensive description of student characteristics, and using online learning algorithms to update student portraits in real time, student portraits can quickly adapt to changes in student learning status.

[0031] Furthermore, matching corresponding art learning content in an art education resource library according to the art student portrait includes:

[0032] Mapping the art student portrait to the knowledge graph corresponding to the art education resource library;

[0033] Using a graph neural network to train the knowledge graph to generate a list of candidate art learning content resources;

[0034] Performing a similarity comparison between the historical learning data of similar students and the candidate art learning content resource list, and using a matrix decomposition method to sort and filter the comparison results;

[0035] The candidate art learning content resource list and the sorting and filtering results are combined to obtain art learning content that matches the art student portrait.

[0036] From the above description, it can be seen that using graph neural networks to train knowledge graphs can better utilize the relationships between knowledge to select a list of candidate art learning content resources, and combine the candidate art learning content resource list and the sorting and filtering results to recommend learning resources that similar learners like but that students may not have yet come into contact with through a hybrid recommendation method, thereby enriching and optimizing personalized learning paths and balancing the personalized needs of learning content and universal laws.

[0037] Furthermore, generating an art education learning path according to the art student portrait and the art learning content includes:

[0038] Establishing a state space and an action space, wherein the state space includes the art student portrait, and the action space includes the art learning content;

[0039] A deep Q network is used to train the state space and the action space to obtain an art education learning path, and a Monte Carlo tree search is used to locally optimize the art education learning path based on student feedback received in real time.

[0040] From the above description, it can be seen that by processing the state space containing the portraits of art students and the action space containing art learning content through the deep Q network, the deep Q network can process high-dimensional state space, improve generalization ability, and can use real-time student feedback to locally optimize the art education learning path to quickly respond to changes in students' needs.

[0041] Further, an art education learning path is generated according to the art student portrait and the art learning content, and then includes:

[0042] Collecting learning effect data of students using the art education learning path;

[0043] Feature extraction is performed on the learning effect data, and multi-dimensional scoring is performed using the extracted feature data, and a learning feedback report for the student is generated based on the multi-dimensional scoring.

[0044] From the above description, it can be seen that by extracting features and scoring multi-dimensionally the learning effect data, the learning path can be dynamically adjusted according to the student's level.

[0045] Please refer to Figure 2 Another embodiment of the present invention provides a terminal for generating an art education learning path based on a large-screen device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the processor implements the various steps of the above-mentioned method for generating an art education learning path based on a large-screen device.

[0046] The above-mentioned method and terminal for generating an art education learning path based on a large-screen device of the present invention are suitable for generating a personalized and flexible art education learning path according to the learning information of students through the large-screen device, and are described below through specific implementation methods:

[0047] Please refer to Figure 1 , Figure 3 and Figure 4 , Embodiment 1 of the present invention is:

[0048] A method for generating an art education learning path based on a large-screen device comprises the following steps:

[0049] S1. Collect art learning data of students on large-screen devices, where the art learning data includes static student data and dynamic interaction data, and use the art learning data to build an art student portrait for the corresponding student.

[0050] S11. Collect students’ art learning data.

[0051] Among them, art learning data includes static student data and dynamic interaction data. Static student data includes historical learning data, and dynamic interaction data includes facial expression data, body movement data and interaction instruction data.

[0052] Specifically, data such as students' facial expressions and body movements are collected through the cameras and sensors of large-screen devices; data on students' interactions with the large screen, including touch operations and voice commands, are analyzed; and historical learning data of students is collected, such as course completion status and work ratings.

[0053] In some embodiments, dynamic interaction data may also include emotional state data and real-time learning behavior obtained by analyzing data collected by cameras and sensors of large-screen devices; static student data may also include basic student information and interests and hobbies.

[0054] S12. Constructing a portrait of an art student. Specifically, the static characteristics and dynamic behaviors of the students are comprehensively considered, and machine learning technology is used to extract features and calculate weights to construct a portrait of the art student.

[0055] S121. Use a deep learning model to extract first feature data of the facial expression data and the body movement data, use a natural language processing model to analyze the interaction instruction data to obtain second feature data, and use time series analysis to process the historical learning data to obtain third feature data.

[0056] Specifically, deep learning models (such as CNN+LSTM) are used to extract students' facial expressions and body movement features; natural language processing technology is applied to analyze students' interactive command data, which includes voice commands and text inputs; and time series analysis methods are used to process students' historical learning data.

[0057] S122: Perform weighted fusion on the first feature data, the second feature data, and the third feature data to obtain a fused feature.

[0058] Specifically, a multimodal feature fusion method with an attention mechanism is used to weight features of different dimensions. The calculation formula for the fused feature vector F is:

[0059] F=∑(w i *f i )

[0060] In the formula, w i represents the weight of the i-th eigenvector, f i represents the i-th eigenvector.

[0061] S123. Perform cluster analysis on the fusion features to obtain the learning type of the student corresponding to the art learning data, wherein the learning type includes a learning preference type, a learning style type, and a student level type, and generate a first student portrait according to the learning type.

[0062] S124. Based on the art learning data of the student acquired in real time, use an online learning algorithm to update the first student portrait in real time to obtain a real-time art student portrait.

[0063] Specifically, we use online learning algorithms, such as Online Gradient Descent, to update student portraits in real time. Update the student portrait model parameters θ t The formula is:

[0064]

[0065] In the formula, θ trepresents the model parameters at time t, η represents the learning rate, L(θ t ) represents the loss function.

[0066] It can be seen that the method for constructing an art student portrait in this embodiment, combining static and dynamic data, can provide a more comprehensive description of student characteristics, can quickly adapt to changes in student learning status, automatically identify and focus on the most relevant characteristics, and improve the accuracy of the portrait. In this way, more accurate personalized recommendations are provided to achieve efficient use of teaching resources.

[0067] S2. Matching corresponding art learning content in an art education resource library according to the art student portrait.

[0068] S21. Construct a knowledge map of art education resource library.

[0069] Specifically, in order to highlight the uniqueness of knowledge graph construction, it is necessary to combine multiple data sources such as art history literature, expert knowledge, online art encyclopedias, etc. to ensure the comprehensiveness and authority of knowledge; design an ontology structure specific to the field of aesthetic education, including core concepts such as art schools, expression techniques, and aesthetic principles; introduce timelines and geographic information to reflect the historical context and regional characteristics of art development; establish connections between different art forms (such as painting, music, and dance) to reflect the comprehensiveness of aesthetic education; add emotional labels to works of art to capture the emotional qualities of art; describe the steps and techniques of artistic creation to help learners understand the creative process; incorporate professional reviews and public feedback to form a multi-dimensional aesthetic evaluation system.

[0070] The construction method of this embodiment is: use natural language processing technology to extract entities and relationships from text materials; use a graph database (such as Neo4j) to store and manage knowledge graphs, where graph nodes correspond to learning concepts, artworks, artists and other data, and edges correspond to conceptual relationships, creative relationships and other data.

[0071] S22. Map the art student portrait to the knowledge graph corresponding to the art education resource library.

[0072] Specifically, the student portrait is mapped to the knowledge graph to find the node most relevant to the student's characteristics. The specific implementation steps are as follows:

[0073] S221. Convert the various features in the student profile (such as interests, skill levels, learning styles, etc.) into numerical vectors, and use word embedding technology (such as Word2Vec) to convert text features into vector representations.

[0074] S222. Apply graph embedding algorithms (such as TransE or Node2Vec) to the knowledge graph to convert the nodes in the graph into low-dimensional vector representations.

[0075] S223. Use cosine similarity or Euclidean distance to calculate the similarity between the student feature vector and the graph node vector.

[0076] S224. Use KNN (K-Nearest Neighbors) algorithm to find the Top-K knowledge graph nodes that are most similar to the student's feature vector.

[0077] S225. Assign weights to the selected nodes according to the similarity scores.

[0078] S226. Based on the edge relationship of the selected node, walk a limited number of steps in the graph to expand the set of related nodes.

[0079] S227. Construct the selected node and its associated nodes into a personalized subgraph as the knowledge interest graph of the student.

[0080] S23, using a graph neural network to train the knowledge graph to generate a list of candidate art learning content resources. Specifically, a graph neural network (such as GraphSAGE) is used to walk on the knowledge graph to generate a list of candidate art learning content resources.

[0081] S24. Perform a similarity comparison between the historical learning data of similar students and the candidate art learning content resource list, and use a matrix decomposition method to sort and filter the comparison results.

[0082] Using matrix decomposition technology, candidate resources are ranked based on the historical learning data of similar students. The specific collaborative filtering steps are as follows:

[0083] S241. Create a matrix R, where R ij Represents the score or interaction intensity of student user i on learning content j.

[0084] S242, Matrix decomposition: Use singular value decomposition (SVD) or non-negative matrix factorization (NMF) to decompose R into two low-dimensional matrices: R≈U*V T , where U represents the user latent factor matrix and V represents the learning content latent factor matrix.

[0085] S243, similar user identification: Based on the U matrix, use cosine similarity to calculate the similarity between users and select K users who are most similar to the target user.

[0086] S244, rating prediction: For each learning content in the candidate resource list, a weighted average is performed based on the ratings of similar users to predict the possible rating of the target user.

[0087] S245. Resource sorting: Sort candidate resources in descending order according to the predicted scores.

[0088] S246, cold start processing: For new users or new learning content, combine content-based recommendation methods, such as using feature information of items to calculate similarity.

[0089] S247. Use incremental learning algorithms, such as Online Alternating Least Squares (Online ALS), to update the matrix decomposition model in real time.

[0090] Through this collaborative filtering method, students can be recommended learning resources that similar learners like but that they may not have yet been exposed to, thereby enriching and optimizing personalized learning paths.

[0091] Matrix factorization technology is used to rank candidate resources based on historical learning data of similar students.

[0092] S25. Combining the candidate art learning content resource list and the sorting and filtering results, obtain art learning content that matches the art student portrait.

[0093] Specifically, the graph-based recommendation results and collaborative filtering results are weighted and fused to obtain the final recommendation list. The formula for fusion scoring is:

[0094] Score(i)=α*ScoreGraph(i)+(1-α)*ScoreCF(i)

[0095] Among them, Score(i) is the final score of resource i, α is the weight coefficient, ScoreGraph(i) is the score calculation method for graph recommendation results, and ScoreCF(i) is the score calculation method for collaborative filtering results.

[0096] The calculation formula of ScoreGraph(i) is:

[0097] ScoreGraph(i)=β·SimNode(i)+(1-β)·PageRank(i);

[0098]

[0099]

[0100] In the formula, β represents the balance factor between node similarity and importance, and its value range is [0,1]; Represents the user interest vector; is the vector representation of resource i; d is the damping coefficient, which is usually taken as 0.85; j∈In(i) is the set of all nodes pointing to node i, and OutDegree(j) represents the out-degree of the calculated node j.

[0101] The calculation formula of ScoreCF(i) is:

[0102]

[0103]

[0104]

[0105] In the formula, u represents the target user; v∈N(u) represents the user in the similar user set; U u , U v Represents the potential factor vector of users u and v; V i represents the latent factor vector of item i; R vi Represents the predicted rating value of user v for item i.

[0106] Therefore, the student portrait construction method in this embodiment can capture the complex knowledge structure in the field of art education by introducing the knowledge graph; by using the graph neural network to process the knowledge graph, it can better utilize the association between knowledge; and by using the hybrid recommendation strategy, it can balance personalized needs and universal laws. In this way, the interpretability of the recommendation results is enhanced, the cold start problem can be effectively alleviated, and the utilization rate of learning resources is improved.

[0107] S3. Generate an art education learning path according to the art student portrait and the art learning content.

[0108] Step S3 specifically includes the following steps: establishing a state space and an action space, wherein the state space includes the art student portrait, and the action space includes the art learning content; using a deep Q network to train the state space and the action space to obtain an art education learning path, and based on the student feedback received in real time, using a Monte Carlo tree search to locally optimize the art education learning path.

[0109] Specifically, define the state space: S = {student portrait, current learning progress, knowledge points mastered}, define the action space: A = {recommend the next learning resource, adjust the difficulty, insert review content}, and design the reward function: R = w1*learning effect + w2*learning efficiency + w3*learning interest, w1, w2, w3 all represent corresponding weights.

[0110] Then, use the Deep Q Network (DQN) to learn the optimal strategy:

[0111] Q(s,a)=Q(s,a)+α[r+γ*max(Qs',a'))-Q(s,a)]

[0112] In the formula, α represents the learning rate, γ represents the discount factor, and s' represents the next state.

[0113] Based on the learned Q-values, the ε-greedy strategy is used to generate a learning path.

[0114] During the learning process, Monte Carlo Tree Search (MCTS) is used to locally optimize the path based on real-time feedback from students.

[0115] Therefore, in the learning path method of this embodiment, the generation of learning paths is regarded as a sequential decision problem, and a deep Q network is used to process high-dimensional state space to improve generalization ability; Monte Carlo tree search is used to achieve real-time dynamic adjustment of the path. In this way, the learning path can be optimized for a long time, with stronger adaptability, and can quickly respond to changes in student needs, and balance exploration and utilization to avoid local optimality.

[0116] In this embodiment, an intelligent assistant with a personified image can also be created to appear on the large screen to guide students to complete learning tasks. It can also analyze students' learning performance and emotional state in real time. Based on the analysis results, the intelligent assistant gives personalized suggestions and dynamically adjusts the learning path.

[0117] The following provides an application scenario of this embodiment:

[0118] In a middle school's art classroom, a smart interactive large screen was installed. Student Xiao Ming walked into the classroom and stood in front of the large screen. The camera recognized Xiao Ming's identity and retrieved his learner portrait. An anthropomorphic smart assistant "Xiao Yi" immediately appeared on the large screen and greeted Xiao Ming.

[0119] Based on Xiao Ming's learner profile, Xiaoyi recommended a course on Impressionist painting to him. After the course started, Monet's Impression, Sunrise was shown on the big screen. Xiaoyi guided Xiao Ming to observe the details of the painting and answered Xiao Ming's questions through voice interaction. When explaining the use of colors, it was detected that Xiao Ming's attention was distracted, so Xiaoyi immediately adjusted the strategy and transformed the abstract color theory into an interactive game.

[0120] Xiao Ming practiced color mixing with his fingers on the big screen, and Xiao Yi analyzed his operation in real time and gave targeted guidance. After completing the practice, Xiao Yi dynamically adjusted the subsequent learning content based on Xiao Ming's performance and added more practical links on the use of colors.

[0121] At the end of the course, Xiao Ming's learning effect was evaluated. It was found that he had made significant progress in color perception, but his composition ability needed to be improved. Based on this result, Xiao Ming's learner profile was updated, and new learning goals and content were set for the next course.

[0122] Throughout the learning process, Xiao Ming maintained a high level of interest and participation. The large-screen smart assistant not only provided a rich audio-visual experience, but also adjusted the teaching strategy according to Xiao Ming's real-time response, truly realizing personalized aesthetic learning.

[0123] Please refer to Figure 3 and Figure 4 , Embodiment 2 of the present invention is:

[0124] The difference between this embodiment and the first embodiment is that, after step S3, the following is further included:

[0125] Collect learning effect data of students using the art education learning path; extract features from the learning effect data, and use the extracted feature data to perform multi-dimensional scoring, and generate a learning feedback report for the students based on the multi-dimensional scoring.

[0126] Specifically, the multi-dimensional scoring includes the following steps:

[0127] (1) Data collection: Collect objective data, including test scores, work scores, completion time, etc.; collect subjective data, including emotional state, participation, self-evaluation, etc.; collect process data, including learning behavior trajectories, interaction data, etc.

[0128] (2) Feature extraction: Use convolutional neural networks to extract visual features of students’ works; apply recurrent neural networks to analyze learning behavior sequences; and use affective computing technology to evaluate students’ emotional states.

[0129] (3) Multi-dimensional scoring: using the features extracted above, adaptive evaluation based on item response theory (IRT); using novelty and practicality scoring models; and multi-criteria evaluation based on expert systems.

[0130] (4) The fuzzy comprehensive evaluation method is used to weight the multi-dimensional scores. The calculation formula of the weighted fusion score E is: E = ∑(q i *S i )

[0131] In the formula, q i represents the weight of the i-th dimension, S i is the score of the i-th dimension.

[0132] (5) Use time series analysis methods, such as the ARIMA model, to assess students’ long-term progress.

[0133] (6) Based on the evaluation results, use natural language generation technology to generate personalized learning feedback reports.

[0134] The evaluation method in this embodiment can fully capture various information in the learning process through multimodal data fusion; can dynamically adjust the evaluation difficulty according to the student's level through adaptive assessment; and can break through the limitations of traditional evaluation through the quantification of creativity and aesthetic ability. In this way, the evaluation results are more comprehensive and objective, which can discover the potential talents of students and provide accurate data support for teaching improvement.

[0135] Please refer to Figure 2 , Embodiment 3 of the present invention is:

[0136] A terminal 1 for generating an art education learning path based on a large-screen device comprises a memory 2, a processor 3 and a computer program stored in the memory 2 and executable on the processor 3. When the processor 3 executes the computer program, each step of a method for generating an art education learning path based on a large-screen device in embodiment one or two is implemented.

[0137] In summary, the present invention provides a method and terminal for generating an art education learning path based on a large-screen device, which collects static student data and dynamic interaction data of students in the large-screen device, and uses the static student data and dynamic interaction data to construct an art student portrait for the corresponding student; according to the art student portrait, the corresponding art learning content is matched in the art education resource library, and the art education learning path is generated according to the art student portrait and the art learning content. Therefore, the interactive data between the student and the large-screen device and the student's own data can be fully utilized to construct an art student portrait, and the learning path is generated by combining the art student portrait and the generated art learning content, ensuring the personalization and high flexibility of the learning path.

[0138] Among them, the present invention provides each student with a customized aesthetic learning path by constructing a detailed learner portrait, effectively improving learning effects and interests. Large-screen devices and smart assistants can also be used to create an immersive learning environment for students to enhance the fun and participation of learning. It is possible to dynamically adjust learning content and strategies based on students' real-time reactions and learning performance to ensure the flexibility and pertinence of the learning process. In some embodiments, various forms of learning resources such as video courses, interactive exercises, and virtual visits can be integrated to enrich the content and form of aesthetic education. Through continuous data collection and analysis, the system continuously optimizes the learning path generation algorithm to achieve long-term improvement in the effect of aesthetic education.

[0139] Compared with the existing technology, the present invention has significant advantages in personalization, interactive experience, real-time adjustment capability and learning resource integration. It can better meet the special needs of aesthetic education and improve students' artistic literacy and learning interest.

[0140] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent transformations made using the contents of the present invention's specification and drawings, or directly or indirectly applied in related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for generating an art education learning path based on a large-screen device, characterized in that: Includes steps: Collecting art learning data of students on large-screen devices, the art learning data including static student data and dynamic interaction data, and using the art learning data to construct an art student portrait for the corresponding student; Matching corresponding art learning content in an art education resource library according to the art student portrait; An art education learning path is generated according to the art student portrait and the art learning content.

2. The method for generating an art education learning path based on a large-screen device according to claim 1, characterized in that: Using the art learning data, an art student portrait is constructed for the corresponding student, including: The static student data includes historical learning data, and the dynamic interaction data includes facial expression data, body movement data and interaction instruction data; Use a deep learning model to extract first feature data of the facial expression data and the body movement data, use a natural language processing model to analyze the interaction instruction data to obtain second feature data, and use time series analysis to process the historical learning data to obtain third feature data; Performing weighted fusion on the first feature data, the second feature data and the third feature data to obtain a fused feature; Performing cluster analysis on the fusion features to obtain the learning type of the student corresponding to the art learning data, the learning type including learning preference type, learning style type and student level type, and generating a first student portrait according to the learning type; According to the art learning data of students acquired in real time, the first student portrait is updated in real time using an online learning algorithm to obtain a real-time art student portrait.

3. The method for generating an art education learning path based on a large-screen device according to claim 2, characterized in that: According to the art student portrait, the corresponding art learning content is matched in the art education resource library, including: Mapping the art student portrait to the knowledge graph corresponding to the art education resource library; Using a graph neural network to train the knowledge graph to generate a list of candidate art learning content resources; Performing a similarity comparison between the historical learning data of similar students and the candidate art learning content resource list, and using a matrix decomposition method to sort and filter the comparison results; The candidate art learning content resource list and the sorting and filtering results are combined to obtain art learning content that matches the art student portrait.

4. The method for generating an art education learning path based on a large-screen device according to claim 1, characterized in that: Generating an art education learning path according to the art student portrait and the art learning content includes: Establishing a state space and an action space, wherein the state space includes the art student portrait, and the action space includes the art learning content; A deep Q network is used to train the state space and the action space to obtain an art education learning path, and a Monte Carlo tree search is used to locally optimize the art education learning path based on student feedback received in real time.

5. The method for generating an art education learning path based on a large-screen device according to claim 1, characterized in that: Generate an art education learning path according to the art student portrait and the art learning content, and then include: Collecting learning effect data of students using the art education learning path; Feature extraction is performed on the learning effect data, and multi-dimensional scoring is performed using the extracted feature data, and a learning feedback report for the student is generated based on the multi-dimensional scoring.

6. A terminal for generating an art education learning path based on a large-screen device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the following steps are implemented: Collecting art learning data of students on large-screen devices, the art learning data including static student data and dynamic interaction data, and using the art learning data to construct an art student portrait for the corresponding student; Matching corresponding art learning content in an art education resource library according to the art student portrait; An art education learning path is generated according to the art student portrait and the art learning content.

7. A terminal for generating art education learning paths based on large-screen devices according to claim 6, characterized in that: Using the art learning data, an art student portrait is constructed for the corresponding student, including: The static student data includes historical learning data, and the dynamic interaction data includes facial expression data, body movement data and interaction instruction data; Use a deep learning model to extract first feature data of the facial expression data and the body movement data, use a natural language processing model to analyze the interaction instruction data to obtain second feature data, and use time series analysis to process the historical learning data to obtain third feature data; Performing weighted fusion on the first feature data, the second feature data and the third feature data to obtain a fused feature; Performing cluster analysis on the fusion features to obtain the learning type of the student corresponding to the art learning data, the learning type including learning preference type, learning style type and student level type, and generating a first student portrait according to the learning type; According to the art learning data of students acquired in real time, the first student portrait is updated in real time using an online learning algorithm to obtain a real-time art student portrait.

8. The terminal for generating art education learning paths based on large-screen devices according to claim 7, characterized in that: According to the art student portrait, the corresponding art learning content is matched in the art education resource library, including: Mapping the art student portrait to the knowledge graph corresponding to the art education resource library; Using a graph neural network to train the knowledge graph to generate a list of candidate art learning content resources; Performing a similarity comparison between the historical learning data of similar students and the candidate art learning content resource list, and using a matrix decomposition method to sort and filter the comparison results; The candidate art learning content resource list and the sorting and filtering results are combined to obtain art learning content that matches the art student portrait.

9. The terminal for generating art education learning paths based on large-screen devices according to claim 6, characterized in that: Generating an art education learning path according to the art student portrait and the art learning content includes: Establishing a state space and an action space, wherein the state space includes the art student portrait, and the action space includes the art learning content; A deep Q network is used to train the state space and the action space to obtain an art education learning path, and a Monte Carlo tree search is used to locally optimize the art education learning path based on student feedback received in real time.

10. The terminal for generating art education learning paths based on large-screen devices according to claim 6, characterized in that: Generate an art education learning path according to the art student portrait and the art learning content, and then include: Collecting learning effect data of students using the art education learning path; Feature extraction is performed on the learning effect data, and multi-dimensional scoring is performed using the extracted feature data, and a learning feedback report for the student is generated based on the multi-dimensional scoring.

Images