Teaching evaluation method and device based on artificial intelligence

Abstract

The invention provides a teaching evaluation method and device based on artificial intelligence, and the method comprises the steps: constructing a subject standard knowledge graph based on a large language model and collected teaching materials, and constructing pre-learning knowledge graphs of different students based on the subject standard knowledge graph; virtual student agents corresponding to different students are constructed based on a multi-modal large model and pre-learning knowledge maps, character types and classroom attention mechanisms of the different students; pre-learning knowledge maps of different students are updated based on an input teacher teaching video through virtual student intelligent agents corresponding to the different students so as to simulate learning behaviors of the different students and obtain post-learning knowledge maps of the different students; and evaluating the teaching quality of the teacher corresponding to the teacher teaching video based on the subject standard knowledge graph, the pre-learning knowledge graph, the post-learning knowledge graph and the learning behavior simulation result of the different students. According to the invention, the efficiency and accuracy of teacher teaching quality evaluation can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of teaching evaluation, and in particular to a teaching evaluation method and device based on artificial intelligence. Background Art

[0002] Pre-service teachers are an important reserve force for the teaching staff. Their ability level in the pre-service education stage largely determines the development of their lifelong abilities. Improving the quality of pre-service teacher training, especially classroom teaching practice ability, is the key to the professional development of pre-service teachers. Establishing a systematic and scientific classroom teaching practice ability evaluation system and realizing accurate diagnosis and improvement of teaching practice ability have become urgent needs to comprehensively improve the professional ability of pre-service teachers.

[0003] The key bottleneck in the process of pre-service teacher training is that the pre-service and post-service evaluation systems are misplaced: the pre-service evaluation targets senior teachers, while the post-service evaluation targets young students. At present, various normal and general universities use educational internships to improve the isolation between teachers and teaching situations in the training of pre-service teachers. However, educational internships still face the following difficulties: 1) During the internship, the teaching time of pre-service teachers is often inconsistent, and they cannot accurately grasp the learning situation of students; 2) During the internship, pre-service teachers need to understand and contact teaching practice from all aspects, which has limited help in improving their subject teaching ability; 3) The academic situation of the internship school is often different from that of the employment school, so pre-service teachers may not learn enough about the academic situation during the internship. Therefore, how to improve the quality of pre-service teacher training has become a key issue and direction of efforts in current teacher education.

[0004] At present, traditional teacher teaching evaluation technology mainly relies on manual methods such as classroom observation and students' after-class feedback to evaluate teachers' teaching quality. The main problem of this method is: due to the lack of data support on factors such as students' personality, learning ability, family and social background, the learning conditions of different students vary greatly. The classroom learning behavior and after-class feedback of some students cannot fully reflect the students' mastery of subject knowledge. Subjective factors have a greater impact, and the manpower cost required for teaching evaluation is high, making it difficult to quickly and accurately evaluate teachers' teaching quality. Summary of the invention

[0005] In view of this, the purpose of the present invention is to provide a teaching evaluation method and device based on artificial intelligence to alleviate the above-mentioned problems existing in traditional teaching evaluation technology.

[0006] In a first aspect, an embodiment of the present invention provides an artificial intelligence-based teaching evaluation method, comprising: constructing a subject standard knowledge graph based on a large language model and collected teaching materials; wherein the subject standard knowledge graph represents knowledge points and the relationship between knowledge points; constructing pre-learning knowledge graphs of different students based on the subject standard knowledge graph; constructing virtual student agents corresponding to different students based on a multimodal large model and pre-learning knowledge graphs, personality types and classroom attention mechanisms of different students; updating the pre-learning knowledge graphs of different students based on the input teacher teaching videos through the virtual student agents corresponding to different students to obtain post-learning knowledge graphs of different students; wherein the virtual student agents simulate the learning behaviors of the corresponding students in the process of updating the pre-learning knowledge graphs of the corresponding students based on the teacher teaching videos; and evaluating the teaching quality of the teacher corresponding to the teacher teaching video based on the subject standard knowledge graph and the pre-learning knowledge graphs, post-learning knowledge graphs and learning behavior simulation results of different students.

[0007] In the second aspect, an embodiment of the present invention also provides an artificial intelligence-based teaching evaluation device, including: a first construction module, used to construct a subject standard knowledge graph based on a large language model and collected teaching materials; wherein the subject standard knowledge graph represents knowledge points and the relationship between knowledge points; a second construction module, used to construct pre-learning knowledge graphs of different students based on the subject standard knowledge graph; a third construction module, used to construct virtual student agents corresponding to different students based on a multimodal large model and pre-learning knowledge graphs, personality types and classroom attention mechanisms of different students; an updating module, used to update the pre-learning knowledge graphs of different students based on the input teacher teaching video through the virtual student agents corresponding to different students, so as to obtain post-learning knowledge graphs of different students; wherein the virtual student agent simulates the learning behavior of the corresponding student in the process of updating the pre-learning knowledge graph of the corresponding student based on the teacher teaching video; an evaluation module, used to evaluate the teaching quality of the teacher corresponding to the teacher teaching video based on the subject standard knowledge graph and the pre-learning knowledge graphs, post-learning knowledge graphs and learning behavior simulation results of different students.

[0008] In a third aspect, an embodiment of the present invention further provides an electronic device, comprising a processor and a memory, wherein the memory stores computer executable instructions that can be executed by the processor, and the processor executes the computer executable instructions to implement the artificial intelligence-based teaching evaluation method described in the first aspect above.

[0009] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the artificial intelligence-based teaching evaluation method described in the first aspect above.

[0010] An embodiment of the present invention provides an artificial intelligence-based teaching evaluation method and device, which constructs a subject standard knowledge graph based on a large language model and collected teaching materials to represent knowledge points and the relationship between knowledge points, constructs pre-learning knowledge graphs of different students based on the subject standard knowledge graph, constructs virtual student agents corresponding to different students based on a multimodal large model and the pre-learning knowledge graphs of different students, personality types and classroom attention mechanisms, updates the pre-learning knowledge graphs of different students based on the input teacher teaching videos through the virtual student agents corresponding to different students to obtain post-learning knowledge graphs of different students, simulates the learning behavior of the corresponding students in the process of updating the pre-learning knowledge graphs of the corresponding students based on the teacher teaching videos, and evaluates the teaching quality of the teacher corresponding to the teacher teaching video based on the subject standard knowledge graph and the pre-learning knowledge graphs of different students, the post-learning knowledge graphs and the learning behavior simulation results. By using the above technology, we can construct knowledge graphs for different students to restore the subject knowledge learning situation of different students as realistically as possible, and construct virtual student intelligent bodies for different students. In combination with teacher teaching videos, we can update the knowledge graphs of different students to simulate the actual learning situation of different students. Then, we can use the learning situation of different students combined with the subject standard knowledge graph to evaluate the teaching quality of teachers, which can improve the efficiency and accuracy of teacher teaching quality evaluation and provide effective support for improving teachers' teaching abilities.

[0011] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0012] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0014] Figure 1 A flowchart of a teaching evaluation method based on artificial intelligence in an embodiment of the present invention; Figure 2 This is an example diagram of the process of constructing a standard knowledge graph in an embodiment of the present invention; Figure 3 This is an example diagram of the process of constructing a personality model in an embodiment of the present invention; Figure 4 This is an example diagram of the process of simulating a learning process for a virtual student in an embodiment of the present invention; Figure 5 This is an example diagram of the process of teaching evaluation in an embodiment of the present invention; Figure 6 Schematic diagram of the structure of a teaching evaluation device based on artificial intelligence in an embodiment of the present invention; Figure 7 The figure is a schematic diagram of the structure of an electronic device in an embodiment of the present invention. DETAILED DESCRIPTION

[0015] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described in combination with the embodiments below. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0016] At present, traditional teacher teaching evaluation technology mainly relies on the manual method of evaluators attending classes and students' feedback after class to evaluate the teaching quality of teachers. The main problem of this method is that due to the lack of data support for factors such as students' personality, learning ability, family and social background, the learning conditions of different students are quite different. The classroom learning behavior and after-class feedback of some students are difficult to fully reflect the students' mastery of subject knowledge. The subjective factors have a great influence, and the human cost required for teaching evaluation is high, making it difficult to quickly and accurately evaluate the teaching quality of teachers. Based on this, the present invention implements a teaching evaluation method and device based on artificial intelligence, which can alleviate the above problems existing in traditional teaching evaluation technology.

[0017] To facilitate understanding of this embodiment, firstly, a teaching evaluation method based on artificial intelligence disclosed in an embodiment of the present invention is introduced in detail, see Figure 1 As shown, the method may include the following steps: Step S102, constructing a subject standard knowledge graph based on the large language model and collected teaching materials.

[0018] Among them, the subject standard knowledge graph can represent knowledge points and the relationships between knowledge points.

[0019] Large language models can be used to extract knowledge from textbooks, teaching aids and other text materials to build a standard knowledge graph.

[0020] Step S104, based on the subject standard knowledge graph, construct the pre-learning knowledge graphs of different students.

[0021] The subject standard knowledge graph can be used as a basis, and different knowledge graphs can be constructed for different students using the subject standard knowledge graph to simulate the knowledge reserves of different students before learning. The nodes and relationships between nodes of the knowledge graphs constructed for different students are generally not exactly the same.

[0022] Step S106, construct virtual student agents corresponding to different students based on the multimodal large model and the pre-learning knowledge graphs, personality types and classroom attention mechanisms of different students.

[0023] The pre-learning knowledge graphs, personality types and classroom attention mechanisms of different students can be configured to the multimodal large model in the form of personalized parameters to obtain multiple different virtual student agents configured with different personalized parameters, so that the virtual student agents can be used to simulate the students' real learning situation in terms of knowledge reserves, personality characteristics and classroom attention concentration.

[0024] Step S108, updating the pre-learning knowledge graphs of different students based on the input teacher teaching video through the virtual student agents corresponding to different students to obtain the post-learning knowledge graphs of different students.

[0025] Among them, the virtual student agent can simulate the learning behavior of the corresponding students in the process of updating the corresponding students' pre-learning knowledge graph based on the teacher's teaching video.

[0026] The teacher's teaching video can be input into each virtual student agent respectively. Each virtual student agent can use the teacher's teaching video to simulate the student's learning behavior and use the corresponding student's learning characteristics to update its own knowledge graph, thereby obtaining the updated knowledge graphs of different virtual student agents.

[0027] Step S110, based on the subject standard knowledge graph and the pre-learning knowledge graphs, post-learning knowledge graphs and learning behavior simulation results of different students, evaluate the teaching quality of the teacher corresponding to the teacher's teaching video.

[0028] The student learning behaviors simulated by different virtual student agents can be combined with the knowledge graphs before and after updating and the standard knowledge graph to evaluate the student learning effects, so as to indirectly reflect the teacher's teaching quality through the student learning effects, thereby realizing the evaluation of the teacher's teaching quality.

[0029] An embodiment of the present invention provides an artificial intelligence-based teaching evaluation method, which constructs a subject standard knowledge graph based on a large language model and collected teaching materials to represent knowledge points and the relationship between knowledge points, constructs pre-learning knowledge graphs of different students based on the subject standard knowledge graph, constructs virtual student agents corresponding to different students based on a multimodal large model and the pre-learning knowledge graphs of different students, personality types and classroom attention mechanisms, updates the pre-learning knowledge graphs of different students based on the input teacher teaching videos through the virtual student agents corresponding to different students to obtain post-learning knowledge graphs of different students, simulates the learning behavior of the corresponding students in the process of updating the pre-learning knowledge graphs of the corresponding students based on the teacher teaching videos, and evaluates the teaching quality of the teacher corresponding to the teacher teaching video based on the subject standard knowledge graph and the pre-learning knowledge graphs of different students, the post-learning knowledge graphs and the learning behavior simulation results. By using the above technology, we can construct knowledge graphs for different students to restore the subject knowledge learning situation of different students as realistically as possible, and construct virtual student intelligent bodies for different students. In combination with teacher teaching videos, we can update the knowledge graphs of different students to simulate the actual learning situation of different students. Then, we can use the learning situation of different students combined with the subject standard knowledge graph to evaluate the teaching quality of teachers, which can improve the efficiency and accuracy of teacher teaching quality evaluation and provide effective support for improving teachers' teaching abilities.

[0030] As a possible implementation, the above step S102 (i.e., constructing a subject standard knowledge graph based on a large language model and collected teaching materials) may include: Step A1, performing text preprocessing on the teaching materials; wherein the text preprocessing may include text cleaning, word segmentation and part-of-speech tagging.

[0031] In step A2, knowledge points are regarded as entities, and relations between knowledge points are regarded as relations between entities. Knowledge extraction is performed on the pre-processed text data through a large language model; wherein knowledge extraction includes entity extraction and relation extraction between entities.

[0032] In step A3, entities are taken as nodes, relationships between entities are taken as edges between nodes, and the knowledge extraction results are used to construct a subject standard knowledge graph.

[0033] See also Figure 2As shown, firstly, the collected textbooks and supplementary teaching materials are preprocessed. The text preprocessing includes text cleaning, word segmentation and part-of-speech tagging. Useless symbols, formatting marks, etc. are removed through text cleaning, and word segmentation and part-of-speech tagging are performed to facilitate the subsequent accurate identification of concepts and entities. Then, the semantic understanding ability of the large language model is used to analyze the sentence structure and reasoning semantic relationships, extract entities (such as theorems, formulas, historical events, etc.) and the relationships between them (such as causal relationships, inclusion relationships, parallel relationships, etc.). Then, the graph database is used to create corresponding nodes for the entities and corresponding edges for the relationships between entities to connect the corresponding nodes. The corresponding nodes can also be deleted according to actual needs, so that the obtained nodes and edges constitute a standard knowledge graph.

[0034] As a possible implementation, the above step S104 (i.e., constructing the pre-learning knowledge graphs of different students based on the subject standard knowledge graph) can be operated in the following two ways: Operation method 1: Analyze the real learning information of different students through a large model to obtain the knowledge points mastered by different students, and modify the subject standard knowledge graph based on the knowledge points mastered by different students to obtain the pre-learning knowledge graph of different students.

[0035] In the case of student information, it can be extracted from the test papers or homework of real students. A large model can be used to analyze student information to obtain students' mastery of knowledge points (such as the attributes and number of knowledge points, the attributes, weights and number of relationships between knowledge points), and align the students' mastery of knowledge points to the specified nodes and edges on the standard knowledge graph, so as to build a personalized knowledge graph. For example, a student correctly solved the coordinates of the vertices of a quadratic function in a math test, but the steps were lengthy, indicating that the student was not proficient in the direct application of the vertex formula, which was reflected in the knowledge graph as the edges (associations) between nodes (quadratic function knowledge points) and nodes (vertex coordinates) were low-level (low weight).

[0036] Operation method 2: Perform random operations corresponding to different students on the subject standard knowledge graph to obtain the pre-learning knowledge graphs of different students. The random operation includes randomly deleting corresponding nodes and / or edges in the subject standard knowledge graph.

[0037] In the absence of student information, some nodes and / or some edges in the standard knowledge graph can be randomly deleted to obtain a personalized knowledge graph to simulate students with different levels of knowledge mastery.

[0038] As a possible implementation, the virtual student agent may include the pre-learning knowledge graph and personality model of the corresponding student; based on this, the above step S106 (i.e., constructing virtual student agents corresponding to different students based on the multimodal large model and the pre-learning knowledge graphs, personality types and classroom attention mechanisms of different students) may include: Step a1: fine-tune the multimodal large model based on the personality types of different students and the classroom attention mechanism to obtain personality models of different students.

[0039] Step a2, based on the pre-learning knowledge graphs and personality models of different students, generate virtual student agents corresponding to different students.

[0040] See also Figure 3 As shown, the construction of personality model mainly includes two parts: feature extraction and model construction; feature extraction can specifically include: extracting typical behavior characteristics (such as interactive behavior, cognitive behavior, emotional behavior, etc.) according to the personality type of students (such as agreeable, conscientious, neurotic, extroverted, open, etc.), extracting attention characteristics (such as the average duration of concentrated attention, the cause of distraction, and the recovery time of attention, etc.) according to the typical behavior characteristics; model construction can specifically include: setting the personality type parameters of the multimodal large model according to the personality type of students and their typical behavior characteristics (such as the category label representing the personality type and the probability distribution of different behaviors, etc.), setting the attention mechanism parameters of the multimodal large model according to the attention characteristics (such as the random time range of the concentration time, the probability relationship function of attention distraction, etc.), the variables of the probability relationship function of attention distraction can include time t, external interference intensity I (such as the intensity of noise, interference from other students, etc.), intrinsic attention stability S (reflecting the degree of concentration of students in the absence of external interference), current knowledge difficulty D (the difficulty of the current course content of students, the more difficult the content is, the more likely it is to cause students to distract their attention), emotional state E (such as anxiety, excitement, etc., which will affect the stability of attention), and the form of the probability relationship function of attention distraction can be: ; , is the amplitude, is the angular frequency, It is the phase; , is the maximum value of the external interference intensity; , is a constant; , is the maximum value of the current knowledge difficulty; , is the maximum value of the emotional state.

[0041] Extraversion: From introversion to extroversion, the scoring range is [0, 1]; students with low scores (introversion) tend to study independently, prefer a quiet environment, and have a longer focus, but may not actively participate in class discussions; students with high scores (extroversion) are more actively involved in class interactions and group discussions, but are easily distracted by external stimuli.

[0042] Agreeableness: From indifferent to friendly, the score range is [0, 1]; students with low scores (indifferent) are more independent and self-oriented, may be insensitive to the opinions of peers or teachers, and have weak collaborative skills; students with high scores (friendly) are more inclined to cooperate with teachers and classmates, are easy to accept suggestions, and are good at teamwork, but may be too dependent on others.

[0043] Conscientiousness: From casual to self-disciplined, the score range is [0, 1]. Students with low scores (casual) tend to procrastinate, lack planning, and have low completion of classroom tasks; students with high scores (self-discipline) have clear goals, good time management skills, and high efficiency in completing classroom tasks.

[0044] Neuroticism: From stable to anxious, the score range is [0, 1]; students with low scores (stable) are emotionally stable and can stay focused in the face of difficulties; students with high scores (anxious) are easily nervous and overly concerned about mistakes, which may lead to reduced learning efficiency, but they may perform better in a low-pressure environment.

[0045] Openness to Experience: From traditional to open, the score range is [0, 1]; students with low scores (traditional) tend to follow the routine, like fixed learning modes, and perform better in structured and clear-rule tasks; students with high scores (open) are more willing to try new things, like interdisciplinary knowledge and creative tasks, are more efficient in innovative tasks, but may ignore details and may be less efficient in structured tasks.

[0046] After setting the personality type and corresponding parameter values ​​of the multimodal large model, according to different parameter settings, the multimodal large model can directly realize the embodiment of personality traits through prompt engineering, that is, directly input personality-related parameters to the multimodal large model for calculation to simulate students' classroom learning behavior. Furthermore, personality traits can be strengthened by fine-tuning the multimodal large model: collect classroom behaviors with specific personality traits, and statistically analyze attention distribution, learning efficiency, etc., and then annotate the obtained data; use open source tools (such as Hugging FaceTransformers) to load the multimodal large model, and use the annotated data for fine-tuning training to obtain a personality model.

[0047] A corresponding pre-learning knowledge graph can be configured for each personality model to construct virtual student agents corresponding to different students. By constructing different types of virtual student agents through a multimodal large model, it is possible to simulate students' behavior in class and their different learning abilities. For example, extroverted students are easily influenced by the outside world, and even actively influence the students around them, causing a loss of attention, which is reflected in the knowledge graph as the loss of knowledge points (i.e. nodes) and / or associations (i.e. edges) in the corresponding time period.

[0048] As a possible implementation, the above-mentioned step S108 (i.e., updating the pre-learning knowledge graphs of different students based on the input teacher teaching video through the virtual student agents corresponding to different students, so as to obtain the post-learning knowledge graphs of different students) may include: inputting the teacher teaching video frame by frame into the virtual student agents corresponding to different students; simulating the learning behavior of the corresponding students by using the personality model of the corresponding students through the virtual student agents corresponding to the corresponding students; extracting key knowledge points from the teacher teaching video by using the simulation results of the learning behavior of the corresponding students through the virtual student agents corresponding to the corresponding students, so as to update the pre-learning knowledge graphs of the corresponding students based on the key knowledge point extraction results of the corresponding students, so as to obtain the post-learning knowledge graphs of different students.

[0049] The personality type determines the classroom behavior (attention level), and the degree of attention determines how much information is actually input to the virtual student agent for reasoning and updating the knowledge graph. Figure 4 As shown, the process of simulating the learning process of a virtual student (i.e., a virtual student agent) can mainly include: Environment initialization: The virtual student loads the corresponding knowledge graph (i.e., the knowledge graph before learning) and personality model, and sets the initial learning state (such as setting the learning progress to 0 and setting the knowledge mastery level to the initial state); Video playback and behavior simulation: Watch the teacher's teaching video, that is, play the teaching video frame by frame to transmit the teaching video to many virtual students. The virtual students record the knowledge contained in the teaching video through the personality model; the virtual students use the personality model to simulate the learning behavior of real students according to the set personality-related parameters during the process of watching the teaching video (such as participating in interaction, listening attentively, being distracted and missing part of the content, etc.), extract key knowledge points from the received teaching content through the personality model, and associate and integrate the extracted key knowledge points with the existing knowledge graph to simulate the student's knowledge absorption process; Knowledge graph update: The virtual student uses the knowledge update capability of the big model to integrate the new knowledge points that have been successfully understood and mastered into its own knowledge graph (such as adding corresponding nodes and updating the relationship between nodes) to update the knowledge graph; the virtual student can also mark the corresponding knowledge points that have errors or difficulties in understanding.

[0050] In the process of virtual student agent simulation learning, the personality model mainly affects the video content obtained by the virtual student agent; specifically, the virtual student agent can obtain complete video content in the video frame during the time period of concentrated attention, and the video content in the video frame obtained by the virtual student agent during the time period of distracted attention is blurred or even missing. The virtual student agent infers the obtained video content to extract key knowledge points, and uses the inference results to update its own knowledge graph.

[0051] As a possible implementation, the above step S110 (i.e., evaluating the teaching quality of the teacher corresponding to the teacher's teaching video based on the subject standard knowledge graph and the pre-learning knowledge graphs, post-learning knowledge graphs and learning behavior simulation results of different students) may include: Step B1, based on the subject standard knowledge graph and the knowledge graphs of different students before and after learning, evaluate the knowledge progress of different students.

[0052] Exemplarily, the knowledge progress assessment result may include the knowledge growth information and knowledge mastery information of the corresponding student; based on this, the operation method of the above step B1 may include the following steps B11 and B12: Step B11, compare the knowledge graph of the same student before learning and the knowledge graph after learning to determine the knowledge growth information of the same student, and obtain the knowledge growth information of different students.

[0053] See also Figure 5 As shown, the knowledge graphs of virtual students before and after learning can be aligned through graph matching technology, and the knowledge growth of the knowledge graph after learning compared with the knowledge graph before learning can be calculated, including the number of newly added knowledge points and changes in knowledge depth (such as the depth of concept understanding and the enhancement of the association between knowledge points).

[0054] Depth of knowledge can be measured by the degree of conceptual understanding and the degree of connection between concepts.

[0055] Concept understanding degree: When the agent obtains relevant information about a new concept, it combines the existing knowledge to form a unique understanding through simple deduction. The similarity between the unique understanding and the standard definition information can be regarded as the degree of understanding of the concept.

[0056] Association: The ratio between the number of associations obtained by the virtual students after watching the teaching video and the number of complete associations in the knowledge graph before learning can be calculated. The size of this ratio can reflect the size of the association. If each association can have a weight, the increase in the weight of the same association after the virtual students watch the teaching video and listen to the lecture compared to the weight of the same association in the knowledge graph before learning can also be calculated. This increase can also reflect the enhancement of the association.

[0057] Step B12, compare the pre-learning knowledge graph and the subject standard knowledge graph of the same student to determine the knowledge mastery information of the same student, and obtain the knowledge mastery information of different students.

[0058] See also Figure 5 As shown, the knowledge graph after learning of the virtual student and the standard knowledge graph can be aligned through graph matching technology to compare the knowledge graph after learning with the standard knowledge graph of the subject. The accuracy of knowledge mastery (such as whether the understanding of knowledge points is correct, specifically whether the positions and attributes of the nodes and edges corresponding to the same knowledge point in the knowledge graph after learning and the standard knowledge graph of the subject are consistent) and completeness (such as whether important knowledge points are omitted, specifically whether the knowledge graph after learning is missing nodes and / or edges corresponding to important knowledge points compared with the standard knowledge graph of the subject) can be evaluated based on the comparison results.

[0059] Step B2, analyzing the learning behavior simulation results of different students to obtain the learning performance data of different students.

[0060] The pre-built statistical model can be used to conduct statistical analysis on the behavioral performance data of virtual students during the learning process (such as the number of interactive participation times, the length of time the students can concentrate, etc.), and the learning effects of the virtual students can be evaluated based on the statistical analysis results, so as to indirectly reflect and evaluate the teaching quality of teachers through the learning effects.

[0061] Step B3, evaluate the teaching quality based on the knowledge progress assessment results and learning performance data of different students.

[0062] After obtaining the knowledge graph comparison results (including knowledge growth and the accuracy and completeness of knowledge mastery) and behavioral performance analysis results (i.e., statistical analysis results of learning performance data) of different virtual students, a detailed teaching evaluation report can be generated using the knowledge graph comparison results and behavioral performance analysis results. The teaching evaluation report includes the advantages of the teacher's teaching (such as vivid and effective teaching methods, clear explanation of knowledge points, etc.), shortcomings (such as some knowledge points are explained too briefly, lack of connection with practical applications, etc.) and targeted improvement suggestions (such as adding case explanations for weak knowledge points, improving teaching methods to improve students' attention, etc.). Teachers can adjust teaching strategies and methods based on the teaching evaluation report to improve their teaching ability.

[0063] For ease of understanding, the implementation of the above-mentioned artificial intelligence-based teaching evaluation method is exemplarily described below by taking a specific application as an example.

[0064] The technical implementation of the above-mentioned AI-based teaching evaluation method mainly consists of the following three parts: (I) Based on the subject knowledge graph, we construct the pre-learning knowledge graphs of different virtual students to represent the subject knowledge reserves of different students, and comprehensively describe the knowledge context and network relationships of students in the learning process of primary school, junior high school and high school. Different virtual students have different levels of mastery of the knowledge graph, and restore the students' subject knowledge learning as realistically as possible. (ii) Construct a personality model of virtual students based on big data and big language models, and simulate the students' real learning situation from multiple aspects such as knowledge reserves, personality traits, and learning ability.

[0065] (III) Through virtual student simulation learning to assist teaching evaluation, the teacher's teaching ability can be evaluated and judged in a targeted manner based on the virtual student's learning situation, thus clarifying the direction for improving the teacher's teaching.

[0066] Based on the above three parts, the technical implementation of the above-mentioned teaching evaluation method based on artificial intelligence mainly includes the following steps: Step 1: Build a standard subject knowledge graph.

[0067] See also Figure 2 As shown, the collected textbooks, teaching aids and other text materials are first preprocessed (including text cleaning, word segmentation and part-of-speech tagging), and then the large language model is used to extract knowledge from the preprocessed data (including extracting entities and the relationships between them), and then a standard knowledge graph is constructed based on the extracted knowledge (including node creation, node connection, node deletion, etc.).

[0068] Step 2: Construction of virtual student agent.

[0069] The virtual student agent consists of two main modules: a knowledge graph that stores knowledge and a multimodal large model that absorbs classroom content and updates the knowledge graph. Based on this, the construction of the virtual student agent includes the construction of the knowledge graph and the construction of the personality model.

[0070] Knowledge graph construction: Based on the standard knowledge graph, the knowledge graph is derived randomly according to the student's situation to simulate the student's knowledge reserve. Specifically, in the absence of student information, some nodes and / or some edges in the standard knowledge graph are randomly deleted to obtain a personalized knowledge graph (i.e., the pre-learning knowledge graph) to simulate students with different levels of knowledge mastery; in the case of student information, a large model is used to analyze the student's mastery of knowledge points based on the real student's test papers or daily homework, and the standard knowledge graph is aligned to construct a personalized knowledge graph.

[0071] Personality model construction: Based on the learning ability distribution, personality characteristics and classroom attention mechanism characteristics of the target student group, the multimodal large model is fine-tuned to generate personality models with different learning characteristics. Specifically, see Figure 3 As shown, according to common personal personality types and their behavioral characteristics as well as key parameters of the attention mechanism (such as the time range of concentration, attention distraction factors, etc.), the personality type and corresponding parameter values ​​of the multimodal large model are set, classroom behaviors with specific personality traits are collected, and the attention distribution, learning efficiency, etc. are counted. The obtained data are annotated, and the multimodal large model is loaded using open source tools and fine-tuned with the annotated data to strengthen the personality traits. After fine-tuning training, the personality model can be obtained.

[0072] After constructing a personalized knowledge graph and personality model, the personalized knowledge graph and personality model can be associated into a virtual student agent to obtain different types of virtual student agents. The virtual student agent can simulate the student's behavior in the classroom, and different types of virtual student agents have different learning abilities.

[0073] Step 3: Teaching evaluation.

[0074] Teaching evaluation includes updating knowledge maps during class, evaluating knowledge progress, and analyzing performance during class.

[0075] Update the knowledge graph during class: The teacher's teaching video is input to the virtual student agent for viewing. The virtual student agent updates its own knowledge graph based on its knowledge graph and learning characteristics. Figure 4As shown, the virtual student agent loads the corresponding knowledge graph and personality model and watches the teacher's teaching video to record the received knowledge through the big model; the virtual student agent simulates the learning behavior of real students in the process of watching the teaching video according to the personality setting parameters, and uses the knowledge updating ability of the big model to extract new knowledge from the received teaching content and integrate it into its own knowledge graph, thereby updating the nodes and edges of its own knowledge graph.

[0076] Assessing Knowledge Progress: See Figure 5 As shown, the updated knowledge graph of the virtual student agent is compared with the knowledge graph before learning to calculate the knowledge growth; at the same time, the updated knowledge graph of the virtual student agent is compared with the standard knowledge graph to evaluate the accuracy and completeness of knowledge mastery.

[0077] Performance analysis during class: see Figure 5 As shown, the behavioral performance data of the virtual student agent during the learning process is statistically analyzed to evaluate the learning effect, and then indirectly evaluate the teaching quality of the teacher. Step 4: Assisted teaching.

[0078] See also Figure 5 As shown, based on the teaching evaluation results (including the behavioral performance of the virtual student intelligent body in the learning process, the knowledge growth of the virtual student intelligent body, and the accuracy and completeness of the virtual student intelligent body's knowledge mastery), a detailed teaching evaluation report (including the teacher's teaching advantages, the teacher's teaching deficiencies, and targeted improvement suggestions) is generated to assist teachers in improving their teaching abilities.

[0079] The advantages of the above-mentioned AI-based teaching evaluation method are mainly reflected in the following aspects: (1) Objective and comprehensive evaluation: Virtual student agents can simulate students of different personality types and learning levels, overcoming the problem of single sample in traditional teacher teaching evaluation methods and more comprehensively reflecting the effect of teachers’ teaching on different student groups. Evaluation based on knowledge graphs and clear evaluation indicators reduces the interference of subjective factors and makes the evaluation results more objective and accurate.

[0080] (2) Personalized assisted teaching: Through the personalized learning feedback of virtual student agents, teachers can gain a deeper understanding of the impact of each teaching link on different types of students, thereby adjusting teaching strategies according to students’ specific needs, achieving personalized teaching, and improving the pertinence and effectiveness of teaching.

[0081] (3) Improving teaching efficiency: Teachers can quickly identify teaching problems based on the learning effect evaluation results of the virtual student intelligent body, avoiding the process of blindly trying to improve teaching methods, saving time and energy, and improving the efficiency of teaching improvement.

[0082] (4) Strong scalability: It can be easily applied to the teaching evaluation of teachers in different subjects and grades. It only needs to adjust the knowledge graph and personality model of the virtual student agent according to the corresponding teaching content and student characteristics. At the same time, with the continuous development of large language model technology, related performance and functions can also be continuously optimized and expanded.

[0083] Based on the above-mentioned teaching evaluation method based on artificial intelligence, the embodiment of the present invention also provides a teaching evaluation device based on artificial intelligence, see Figure 6 As shown, the device may include the following modules: The first construction module 702 is used to construct a subject standard knowledge graph based on the large language model and the collected teaching materials; wherein the subject standard knowledge graph represents knowledge points and the association relationship between knowledge points; The second construction module 704 is used to construct the pre-learning knowledge graphs of different students based on the subject standard knowledge graph; The third construction module 706 is used to construct virtual student agents corresponding to different students based on the multimodal large model and the pre-learning knowledge graphs, personality types and classroom attention mechanisms of different students; The updating module 708 is used to update the pre-learning knowledge graphs of different students based on the input teacher teaching video through the virtual student agents corresponding to different students, so as to obtain the post-learning knowledge graphs of different students; wherein the virtual student agents simulate the learning behavior of the corresponding students in the process of updating the pre-learning knowledge graphs of the corresponding students based on the teacher teaching video; The evaluation module 710 is used to evaluate the teaching quality of the teacher corresponding to the teacher's teaching video based on the subject standard knowledge graph and the pre-learning knowledge graphs, post-learning knowledge graphs and learning behavior simulation results of different students.

[0084] By adopting the above-mentioned artificial intelligence-based teaching evaluation device, knowledge graphs of different students can be constructed to restore the subject knowledge learning situation of different students as realistically as possible, and virtual student intelligent bodies of different students can be constructed to update the knowledge graphs of different students in combination with teacher teaching videos to simulate the actual learning situation of different students. Then, the learning situation of different students can be combined with the subject standard knowledge graph to evaluate the teaching quality of teachers, which can improve the efficiency and accuracy of teacher teaching quality evaluation and provide effective support for improving teachers' teaching ability.

[0085] The artificial intelligence-based teaching evaluation device provided in the embodiment of the present invention has the same implementation principle and technical effects as the aforementioned artificial intelligence-based teaching evaluation method embodiment. For the sake of brief description, for matters not mentioned in the embodiment of the artificial intelligence-based teaching evaluation device, reference may be made to the corresponding contents in the aforementioned artificial intelligence-based teaching evaluation method embodiment.

[0086] The embodiment of the present invention further provides an electronic device, such as Figure 7 As shown, it is a schematic diagram of the structure of the electronic device, wherein the electronic device includes a processor 71 and a memory 70, the memory 70 stores computer executable instructions that can be executed by the processor 71, and the processor 71 executes the computer executable instructions to implement the above-mentioned artificial intelligence-based teaching evaluation method.

[0087] exist Figure 7 In the illustrated embodiment, the electronic device further includes a bus 72 and a communication interface 73 , wherein the processor 71 , the communication interface 73 and the memory 70 are connected via the bus 72 .

[0088] Among them, the memory 70 may include a high-speed random access memory (RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk storage. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 73 (which can be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. can be used. The bus 72 can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The bus 72 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 7 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0089] The processor 71 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit in the processor 71 or the instruction in the form of software. The above processor 71 can be a general processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The general processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in the embodiment of the present invention can be directly embodied as a hardware decoding processor to execute, or it can be executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor 71 reads the information in the memory and combines its hardware to complete the steps of the teaching evaluation method based on artificial intelligence in the aforementioned embodiment.

[0090] An embodiment of the present invention also provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the above-mentioned artificial intelligence-based teaching evaluation method. The specific implementation can be found in the aforementioned method embodiment, which will not be repeated here.

[0091] The computer program product of the artificial intelligence-based teaching evaluation method, device and electronic device provided in the embodiments of the present invention includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods described in the previous method embodiments. The specific implementation can be found in the method embodiments, which will not be repeated here.

[0092] Unless otherwise specifically stated, the relative steps, numerical expressions and values ​​of the components and steps set forth in these embodiments do not limit the scope of the present invention.

[0093] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that can be executed by a processor. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., various media that can store program codes.

[0094] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0095] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A teaching evaluation method based on artificial intelligence, characterized in that: include: Based on the large language model and the collected teaching materials, a subject standard knowledge graph is constructed; wherein the subject standard knowledge graph represents knowledge points and the relationship between knowledge points; Based on the subject standard knowledge graph, construct pre-learning knowledge graphs of different students; Based on the multimodal large model and different students' pre-learning knowledge graphs, personality types and classroom attention mechanisms, we build virtual student agents corresponding to different students. The virtual student agents corresponding to different students update the pre-learning knowledge graphs of different students based on the input teacher teaching videos to obtain the post-learning knowledge graphs of different students; wherein the virtual student agents simulate the learning behaviors of the corresponding students in the process of updating the pre-learning knowledge graphs of the corresponding students based on the teacher teaching videos; Based on the standard knowledge graph of the subject and the pre-learning knowledge graphs, post-learning knowledge graphs and learning behavior simulation results of different students, the teaching quality of the teacher corresponding to the teacher's teaching video is evaluated.

2. The teaching evaluation method based on artificial intelligence according to claim 1, characterized in that: The virtual student agent includes the pre-learning knowledge graph and personality model of the corresponding students; based on the multimodal large model and the pre-learning knowledge graph, personality type and classroom attention mechanism of different students, the virtual student agents corresponding to different students are constructed, including: Fine-tuning the multimodal large model based on the personality types and classroom attention mechanisms of different students to obtain personality models of different students; Based on the pre-learning knowledge graphs and personality models of different students, virtual student agents corresponding to different students are generated.

3. The teaching evaluation method based on artificial intelligence according to claim 2 is characterized in that: The virtual student agents corresponding to different students update the pre-learning knowledge graphs of different students based on the input teacher teaching videos to obtain the post-learning knowledge graphs of different students, including: Input the teacher teaching video frame by frame into the virtual student agents corresponding to different students; The learning behavior of the corresponding student is simulated by using the corresponding student's personality model through the corresponding virtual student agent; The key knowledge points of the teacher's teaching video are extracted by the virtual student agent corresponding to the corresponding student using the learning behavior simulation results of the corresponding student, so as to update the pre-learning knowledge graph of the corresponding student based on the key knowledge point extraction results of the corresponding student, and obtain the post-learning knowledge graphs of different students.

4. The teaching evaluation method based on artificial intelligence according to claim 2 is characterized in that: Based on the subject standard knowledge graph, construct the pre-learning knowledge graphs of different students, including: The real learning information of different students is analyzed through a large model to obtain the knowledge points mastered by different students, and the subject standard knowledge graph is modified based on the knowledge points mastered by different students to obtain the pre-learning knowledge graphs of different students; or, random operations corresponding to different students are performed on the subject standard knowledge graph to obtain the pre-learning knowledge graphs of different students, and the random operations include randomly deleting corresponding nodes and / or edges in the subject standard knowledge graph.

5. The teaching evaluation method based on artificial intelligence according to claim 1, characterized in that: Based on the subject standard knowledge graph and the pre-learning knowledge graphs, post-learning knowledge graphs and learning behavior simulation results of different students, the teaching quality of the teacher corresponding to the teacher's teaching video is evaluated, including: Based on the subject standard knowledge graph and the pre-learning knowledge graphs and post-learning knowledge graphs of different students, evaluating the knowledge progress of different students; Analyze the learning behavior simulation results of different students to obtain the learning performance data of different students; The teaching quality is evaluated based on the knowledge progress assessment results and learning performance data of different students.

6. The teaching evaluation method based on artificial intelligence according to claim 5 is characterized in that: The knowledge progress assessment results include the knowledge growth information and knowledge mastery information of the corresponding students; Based on the subject standard knowledge graph and the pre-learning knowledge graphs and post-learning knowledge graphs of different students, the knowledge progress of different students is evaluated, including: Compare the knowledge graph of the same student before learning and the knowledge graph after learning to determine the knowledge growth information of the same student, and obtain the knowledge growth information of different students; The pre-learning knowledge graph of the same student and the standard knowledge graph of the subject are compared to determine the knowledge mastery information of the same student, thereby obtaining the knowledge mastery information of different students.

7. The teaching evaluation method based on artificial intelligence according to claim 1, characterized in that: Based on the large language model and collected teaching materials, we build a subject standard knowledge graph, including: Performing text preprocessing on the teaching materials; wherein the text preprocessing includes text cleaning, word segmentation and part-of-speech tagging; Taking knowledge points as entities and relations between knowledge points as relations between entities, knowledge extraction is performed on the pre-processed text data through the large language model; wherein the knowledge extraction includes entity extraction and relation extraction between entities; The entities are taken as nodes, the relationships between entities are taken as edges between nodes, and the knowledge extraction results are used to construct the subject standard knowledge graph.

8. A teaching evaluation device based on artificial intelligence, characterized in that: include: The first construction module is used to construct a subject standard knowledge graph based on a large language model and collected teaching materials; wherein the subject standard knowledge graph represents knowledge points and the relationship between knowledge points; The second construction module is used to construct the pre-learning knowledge graphs of different students based on the subject standard knowledge graph; The third building module is used to build virtual student agents corresponding to different students based on the multimodal large model and the pre-learning knowledge graphs, personality types and classroom attention mechanisms of different students; An updating module, used for updating the pre-learning knowledge graphs of different students based on the input teacher teaching video through the virtual student agents corresponding to different students, so as to obtain the post-learning knowledge graphs of different students; wherein the virtual student agents simulate the learning behavior of the corresponding students in the process of updating the pre-learning knowledge graphs of the corresponding students based on the teacher teaching video; An evaluation module is used to evaluate the teaching quality of the teacher corresponding to the teacher's teaching video based on the subject standard knowledge graph and the pre-learning knowledge graphs, post-learning knowledge graphs and learning behavior simulation results of different students.

9. An electronic device, characterized in that: It comprises a processor and a memory, wherein the memory stores computer executable instructions that can be executed by the processor, and the processor executes the computer executable instructions to implement the artificial intelligence-based teaching evaluation method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions. When the computer-executable instructions are called and executed by the processor, the computer-executable instructions prompt the processor to implement the artificial intelligence-based teaching evaluation method described in any one of claims 1 to 7.

Images