Methods, equipment, devices, and media for evaluating instructional content.

Abstract

This application provides a method, apparatus, device, and medium for evaluating classes. This application belongs to the technical field of teaching evaluation. The method includes the steps of: acquiring class data and class objective information for a class to be evaluated; converting the class data into input text; receiving an objective analysis command for the class to be evaluated; inputting the input text and the class objective information into a predetermined model, the predetermined model processing the input text and the class objective information in response to the objective analysis command, and outputting an evaluation result for the class to be evaluated. By employing this technical solution, the objective of performing question-and-answer dialogue analysis tasks, class interaction analysis tasks, and curriculum standard implementation analysis tasks on a class without artificial listening can be achieved, improving the efficiency of class evaluation. At the same time, since this solution is a class evaluation based on class objective information, objectivity and accuracy of the evaluation results are ensured.

Description

Technical Field

[0001] This application belongs to the technical field of teaching evaluation, and specifically relates to a teaching evaluation method, device, device and medium.

Background Art

[0002] In the teaching process of teachers, through teaching interaction behaviors (such as teaching questions), the key points and difficult points of teaching materials can be focused on, and students can be encouraged to deepen their understanding of knowledge. In the teaching interaction process, inspiring questions related to teaching goals are very important in the teaching process. Therefore, in teaching observation, it is very important to evaluate the teaching of teachers.

[0003] In related technologies, regarding the evaluation of teachers' teaching, in many cases, each school designs a lecture questionnaire or scale based on relevant theories before teaching, and organizes it online or offline for teachers or lecturers specialized in teaching observation to attend. Finally, the teacher or lecturer in charge of teaching observation summarizes, discusses, and inducts the teaching interaction behaviors of the teacher based on the teaching materials collected from the teaching scenario, and proposes the current teaching.

[0004] However, in the current teaching evaluation method, mainly, specialized teaching observation teachers conduct lecture evaluations manually, which takes a long time. At the same time, different observation teachers may have different viewpoints on the teaching interaction behaviors of the same teacher. Therefore, when using related technologies to conduct teaching evaluations, there are problems such as low evaluation efficiency and high subjectivity of evaluation conclusions.

Summary of the Invention

[0005] The object of the embodiments of this application is to provide a method, apparatus, device, and medium for evaluating classes that can solve the problems of low evaluation efficiency and high subjectivity of evaluation conclusions when evaluating class interaction behavior using related technologies. The objective is to achieve the goal of performing question-and-answer dialogue analysis tasks, class interaction analysis tasks, and curriculum standard implementation analysis tasks on a class without artificial listening, by converting class data of a class to be evaluated into input text, receiving an objective analysis command for the class to be evaluated, and inputting the input text and class objective information into the predetermined model, so that the predetermined model processes the input target text and the class objective information in response to the objective analysis command and outputs the evaluation results of the class to be evaluated. This improves the efficiency of class evaluation, and at the same time, since the class evaluation is performed based on the input target text converted from the class objective information and class data, the objectivity and accuracy of the evaluation results are ensured.

[0006] In the first embodiment, the method for evaluating lessons according to the embodiment of this application is: A step of obtaining lesson data and lesson objective information for a lesson to be evaluated, wherein the lesson data includes audio data and / or video data, and the lesson objective information includes teaching objective information and curriculum standard information. The steps include converting the aforementioned lesson data into input text, A step of receiving an objective analysis command for the subject class to be evaluated, wherein the objective analysis command includes a question-and-answer dialogue analysis command, a class interaction analysis command, and a curriculum standard implementation analysis command, The steps include inputting the input target text and the lesson objective information into a predetermined model, the predetermined model processing the input target text and the lesson objective information in accordance with the objective analysis command, and outputting the evaluation result of the lesson to be evaluated, wherein the predetermined model is obtained by using an educational scene corpus as a training sample, using a question-and-answer dialogue analysis main task module, a lesson interaction analysis main task module, and a curriculum standard implementation analysis main task module as the main tasks of collaborative training, and conducting supervised fine-tuning training, the question-and-answer dialogue analysis main task module is used to respond to the question-and-answer dialogue analysis command, the lesson interaction analysis main task module is used to respond to the lesson interaction analysis command, and the curriculum standard implementation analysis main task module is used to respond to the curriculum standard implementation analysis command.

[0007] In this solution, the objective of performing question-and-answer dialogue analysis tasks, lesson interaction analysis tasks, and curriculum standard implementation analysis tasks on a lesson without artificial listening is achieved. This is accomplished by converting lesson data of a lesson to be evaluated into input text, receiving objective analysis commands for the lesson to be evaluated, inputting the input text and lesson objective information into a predetermined model, and having the predetermined model process the input text and lesson objective information in response to the objective analysis commands and output the evaluation results for the lesson to be evaluated. This improves the efficiency of lesson evaluation, and at the same time, since this solution is based on lesson objective information and input text converted from lesson data, it ensures the objectivity and accuracy of the evaluation results.

[0008] In one embodiment, the steps of inputting the input text and the lesson objective information into a predetermined model, the predetermined model processing the input text and the lesson objective information in accordance with the objective analysis command, and outputting the evaluation results of the lesson to be evaluated are as follows: If it is confirmed that the received objective analysis command is a question-and-answer dialogue analysis command, the input target text and the teaching objective information are input into a predetermined model, the predetermined model responds to the question-and-answer dialogue analysis command, executes the question-and-answer dialogue analysis main task module based on the input target text and the teaching objective information, and outputs the evaluation results of the evaluation target lesson in the question-and-answer dialogue analysis main task module. If it is confirmed that the received objective analysis command is a lesson interaction analysis command, the input target text and the teaching objective information are input into a predetermined model, the predetermined model responds to the lesson interaction analysis command, executes the lesson interaction analysis main task module based on the input target text and the teaching objective information, and outputs the evaluation results of the lesson to be evaluated in the lesson interaction analysis main task module. The process includes the steps of: confirming that the received objective analysis command is a curriculum standard implementation analysis command, inputting the input target text and the curriculum standard information into a predetermined model; the predetermined model responding to the curriculum standard implementation analysis command, executing the curriculum standard implementation analysis main task module based on the input target text and the curriculum standard information, and outputting the evaluation results of the evaluation target class in the curriculum standard implementation analysis main task module.

[0009] The beneficial effects of this solution are as follows: In response to different types of objective analysis commands received, the input text and lesson objective information are input into a predetermined model. The predetermined model responds to different objective analysis commands by executing the corresponding main task module based on the curriculum standard information or teaching objective information within the lesson objective information and the input text. The evaluation results of the lesson to be evaluated are output in the correspondingly selected main task module. This achieves the objective of performing evaluations corresponding to the lesson to be evaluated according to the evaluation tasks in different types of main task modules, improving the objectivity of the evaluation of lesson interaction behaviors and the user experience when viewing the evaluation results.

[0010] In one embodiment, when it is confirmed that the received objective analysis command is a question-and-answer dialogue analysis command, the input target text and the teaching objective information are input into a predetermined model, the predetermined model responds to the question-and-answer dialogue analysis command, executes the question-and-answer dialogue analysis main task module based on the input target text and the teaching objective information, and outputs the evaluation result of the evaluation target lesson in the question-and-answer dialogue analysis main task module, If it is confirmed that the received objective analysis command is a question-and-answer dialogue analysis command, the predetermined model is confirmed to respond to the question-and-answer dialogue analysis command, and the predetermined model is set to wait for the execution of objective evaluation tasks in each type of evaluation dimension in the question-and-answer dialogue analysis main task module. The steps to determine the objective evaluation task, The process includes the steps of inputting the input text and the teaching objective information into a predetermined model, having the question-and-answer dialogue analysis main task module perform the objective evaluation task based on the input text and the teaching objective information, and outputting the evaluation results of the evaluation target lesson.

[0011] The beneficial effects of this solution are as follows: After confirming that a question-and-answer dialogue analysis command has been received and that a predetermined model will respond to the command, the predetermined model enters a state of waiting for the execution of the question-and-answer dialogue analysis main task module. In this case, the predetermined model determines the target evaluation task for each type of evaluation dimension in the question-and-answer dialogue analysis main task module, and by executing the target evaluation task based on the input text and the teaching objective information, the effect of analyzing a teacher's questions during a lesson based on different evaluation dimensions can be achieved.

[0012] In one embodiment, the main task module for question-and-answer dialogue analysis includes an overall evaluation subtask module and a local evaluation subtask module. The aforementioned step of determining the objective evaluation task is, A step of obtaining a first type of evaluation prompt command, the first type of evaluation prompt command is used to prompt a predetermined model to output an evaluation result of the lesson under evaluation according to an evaluation dimension of question effectiveness analysis, the evaluation dimension of question effectiveness analysis includes an evaluation of the effectiveness of the teacher's questions in the lesson under evaluation, A step in which, in response to the first type of evaluation prompt command, it is determined that a first overall subtask in the overall evaluation subtask module and a first local subtask in the local evaluation subtask module jointly constitute a target evaluation task, wherein the first overall subtask is used to overall evaluate the overall effectiveness of all of the teacher's questions in the lesson being evaluated, and the first local subtask is used to specifically determine the effectiveness of each of the teacher's questions in the lesson being evaluated. and / or, A step of obtaining a second type of evaluation prompt command, the second type of evaluation prompt command being used to prompt a predetermined model to output an evaluation result of the lesson being evaluated according to an evaluation dimension of knowledge objectives, the evaluation dimension of knowledge objectives including an evaluation of the relevance between the teacher's questions and the teaching objectives in the lesson being evaluated, A step in which, in response to the second type of evaluation prompt command, it is determined that the second overall subtask in the overall evaluation subtask module and the second local subtask in the local evaluation subtask module jointly constitute an objective evaluation task, wherein the second overall subtask is used to evaluate the overall pre-set knowledge objectives in the lesson being evaluated, and the second local subtask is used to specifically evaluate the degree of correlation between the teacher's questions and the teaching objectives in the lesson being evaluated. and / or, A step of obtaining a third type of evaluation prompt command as the target evaluation task, wherein the third type of evaluation prompt command is used to prompt a predetermined model to output an evaluation result of the lesson to be evaluated according to an evaluation dimension of question avoidance, the evaluation dimension of question avoidance includes a step of identifying teacher avoidable questions in the lesson to be evaluated, A step in which, in response to the third type of evaluation prompt command, it is determined that a first local subtask and a second local subtask in the local evaluation subtask module jointly constitute an objective evaluation task, wherein the first local subtask is used to specifically determine the effectiveness of each question asked by the teacher in the lesson being evaluated, and the second local subtask specifically evaluates the degree of correlation between the teacher's questions and the teaching objectives in the lesson being evaluated. and / or, A step of obtaining a fourth type of evaluation prompt command as the target evaluation task, wherein the fourth type of evaluation prompt command is used to prompt a predetermined model to output an evaluation result of the lesson to be evaluated according to an evaluation dimension of question optimization, the evaluation dimension of question optimization includes finding teacher-optimizable questions in the lesson to be evaluated, In response to the fourth type of evaluation prompt command, the step of determining that a third global subtask in the global evaluation subtask module and a third local subtask in the local evaluation subtask module jointly constitute a target evaluation task, wherein the third global subtask is used to propose an overall optimization of the teacher's questions in the lesson being evaluated, and the third local subtask is used to optimize and rewrite the teacher's questions in the lesson being evaluated.

[0013] The beneficial effects of this solution are as follows: By simultaneously or selectively evaluating the content of the question-and-answer dialogue in the target lesson based on different evaluation dimensions and different evaluation ranges, it is possible to evaluate the teacher's questions in the target lesson in different dimensions, assuming the use of a unified model. This avoids the problem in related technologies where training results in different dimensions become isolated from each other because it is necessary to train each evaluation dimension using lesson data separately. This more effectively reduces the amount of iterative model training work, increases the relevance between training tasks in different evaluation dimensions, and further improves the objectivity of lesson evaluation.

[0014] In one embodiment, the main task module for question-and-answer dialogue analysis includes an overall evaluation subtask module and a local evaluation subtask module, and the predetermined model, in the process of performing supervised fine-tuning training on the main task module for question-and-answer dialogue analysis, processes the lesson text containing the teaching objectives and teacher questions in the educational scene corpus according to the overall evaluation subtask module and the local evaluation subtask module, respectively, and outputs an evaluation result for the lesson text.

[0015] The beneficial effects of this solution are as follows: By providing a question-and-answer dialogue analysis main task module that includes an overall evaluation subtask module and a local evaluation subtask module, and by providing supervised training to the overall evaluation subtask module and the local evaluation subtask module, respectively, based on an educational scene corpus, the objective of performing overall and specific evaluations of teachers' questions and answers can be achieved, and the accuracy of the evaluation results output by the predetermined model can be improved.

[0016] In one embodiment, the overall assessment subtask module includes a first overall subtask, a second overall subtask, and a third overall subtask, the first overall subtask being used in supervised fine-tuning training to overall assess the overall effectiveness of all of the teacher's questions in the lesson text; the second overall subtask being used in supervised fine-tuning training to overall assess the pre-set knowledge objectives in the lesson text; and the third overall subtask being used in supervised fine-tuning training to propose an overall optimization of the teacher's questions in the lesson text. The local assessment subtask module includes a first local subtask, a second local subtask, and a third local subtask, the first local subtask being used in supervised fine-tuning training to specifically determine the effectiveness of each of the teacher's questions in the lesson text; the second local subtask being used in supervised fine-tuning training to specifically assess the correlation between the teacher's questions and the teaching objectives in the lesson text; and the third local subtask being used in supervised fine-tuning training to optimize and rewrite the teacher's questions in the lesson text.

[0017] The beneficial effects of this solution are as follows: By providing an overall evaluation subtask module containing three overall subtasks and a local evaluation subtask module containing three local subtasks, supervised training is jointly conducted for a total of six subtasks (three overall subtasks and three local subtasks). In the training process, the first overall subtask and the first local subtask collaborate to produce evaluation results in the dimension of question effectiveness analysis, the second overall subtask and the second local subtask collaborate to produce evaluation results in the dimension of knowledge objectives, and the third overall subtask and the third local The subtasks collaborate to output evaluation results in the dimension of question optimization, and simultaneously, the first and second local subtasks collaborate to output evaluation results in the dimension of question avoidance. In other words, all subtasks performed collaboratively in the overall evaluation subtask module and the local evaluation subtask module can output evaluation results of teachers' questions in the evaluated lesson according to four evaluation dimensions: question effectiveness analysis, knowledge objectives, question avoidance, and question optimization. This ensures objectivity in lesson evaluation and further improves the comprehensiveness of the analysis of teachers' questions during lessons.

[0018] In one embodiment, the step of acquiring lesson data and lesson objective information for the lesson to be evaluated is: Steps to obtain lesson data for the course to be evaluated, The process includes the steps of querying a pre-configured database for lesson objective information that matches the lesson data, or inputting the lesson data into a pre-configured chapter matching model and causing the chapter matching model to output lesson objective information that matches the lesson data.

[0019] The beneficial effects of this solution are as follows. Query the lesson objective information that matches the lesson data in a pre-set database, or input the lesson data into a pre-set chapter matching model, and output the lesson objective information that matches the lesson data in the chapter matching model, thereby improving the determination efficiency of the lesson objective information.

[0020] In one embodiment, the step of converting the lesson data into input target text includes: Converting the lesson data into text to obtain lesson text; Extracting the teacher's question text from the lesson text; Using the lesson text and the teacher's question text as the input target text.

[0021] The beneficial effects of this solution are as follows. By converting the lesson data into text to obtain lesson text, extracting the teacher's question text from the lesson text, and using the lesson text and the teacher's question text as the input target text, it is beneficial to perform specific analysis and overall analysis combining context for the teacher's questions respectively later, and the comprehensiveness of the input target text is improved.

[0022] In one embodiment, the step of extracting the teacher's question text from the lesson text includes: Identifying the dialogue character with the most dialogue times in the lesson text; Determining the dialogue character with the most dialogue times as the teacher character; Extracting all the dialogue contents of the teacher character; Screening the teacher's question text from all the dialogue contents.

[0023] The beneficial effects of this solution are as follows: By determining the dialogue character with the most frequent dialogues in the lesson text as the teacher character and screening the teacher's question text within the dialogue content, the step of extracting the teacher's question text can be simplified, improving the efficiency and accuracy of the teacher's question text extraction.

[0024] In one embodiment, the step of screening the teacher's question text from all the dialogue content is: The steps include identifying the basic question sentences from all of the above dialogue content, The process includes the step of performing a noise reduction operation on the aforementioned basic question sentence to obtain the teacher's question text.

[0025] The beneficial effects of this solution are as follows: By identifying basic question sentences from all of the teacher's dialogue content, performing noise reduction operations on these basic question sentences, and obtaining the teacher's question text, the objective of further refining the question text can be achieved, and it is possible to avoid the basic question sentences later influencing the evaluation results of the lesson being evaluated.

[0026] In the second embodiment, the lesson evaluation device according to the embodiment of this application is Used to acquire lesson data and lesson objective information for the lessons to be evaluated, the lesson data includes audio data and / or video data, and the lesson objective information includes a data acquisition module that includes teaching objective information and curriculum standard information. A data conversion module for converting the aforementioned lesson data into input text, Used to receive objective analysis commands for the aforementioned evaluation subject class, the objective analysis command includes a command receiving module that includes a question-and-answer dialogue analysis command, a class interaction analysis command, and a curriculum standard implementation analysis command, The input text and the lesson objective information are input into a predetermined model, which processes the input text and the lesson objective information in accordance with the objective analysis command and is used to output the evaluation results of the lesson to be evaluated. The predetermined model is obtained by using an educational scene corpus as a training sample, using a question-and-answer dialogue analysis main task module, a lesson interaction analysis main task module, and a curriculum standard implementation analysis main task module as the main tasks of collaborative training, and conducting supervised fine-tuning training. The question-and-answer dialogue analysis main task module is used to respond to the question-and-answer dialogue analysis command, the lesson interaction analysis main task module is used to respond to the lesson interaction analysis command, and the curriculum standard implementation analysis main task module is used to respond to the curriculum standard implementation analysis command.

[0027] In this solution, the objective of performing question-and-answer dialogue analysis tasks, lesson interaction analysis tasks, and curriculum standard implementation analysis tasks on a lesson without artificial listening is achieved. This is accomplished by converting lesson data of a lesson to be evaluated into input text, receiving objective analysis commands for the lesson to be evaluated, inputting the input text and lesson objective information into a predetermined model, and having the predetermined model process the input text and lesson objective information in response to the objective analysis commands and output the evaluation results for the lesson to be evaluated. This improves the efficiency of lesson evaluation, and at the same time, since this solution is based on lesson objective information and input text converted from lesson data, it ensures the objectivity and accuracy of the evaluation results.

[0028] In a third aspect, an embodiment of the present application provides an electronic device comprising a processor, a memory, and a program or instruction stored in the memory and executable by the processor, wherein when the program or instruction is executed by the processor, the steps of the lesson evaluation method described in the first aspect are realized.

[0029] In a fourth aspect, an embodiment of the present application provides a readable storage medium in which a program or instruction is stored, and when the program or instruction is executed by a processor, the steps of the lesson evaluation method described in the first aspect are realized.

[0030] In a fifth aspect, an embodiment of the present application provides a chip comprising a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to execute a program or instruction and to implement the method according to the first aspect.

[0031] In the embodiments of this application, lesson data and lesson objective information for a lesson to be evaluated are acquired, the lesson data includes audio data and / or video data, the lesson objective information includes teaching objective information and curriculum standard information, the lesson data is converted into input text, an objective analysis command for the lesson to be evaluated is received, the objective analysis command includes a question-and-answer dialogue analysis command, a lesson interaction analysis command and a curriculum standard implementation analysis command, the input text and the lesson objective information are input into a predetermined model, the predetermined model processes the input text and the lesson objective information in accordance with the objective analysis command and outputs the evaluation result for the lesson to be evaluated, and the predetermined model This was achieved by using an educational scene corpus as a training sample, and by using the Q&A dialogue analysis main task module, the classroom interaction analysis main task module, and the curriculum standard implementation analysis main task module as the main tasks of collaborative training, with supervised fine-tuning training being conducted. The Q&A dialogue analysis main task module is used to respond to the Q&A dialogue analysis command, the classroom interaction analysis main task module is used to respond to the classroom interaction analysis command, and the curriculum standard implementation analysis main task module is used to respond to the curriculum standard implementation analysis command. The above-described method for evaluating lessons solves the problems of low efficiency and high subjectivity in evaluation conclusions when evaluating lesson interaction behavior using related technologies. By converting lesson data of the lesson to be evaluated into input text, receiving objective analysis commands for the lesson to be evaluated, inputting the input text and lesson objective information into a predetermined model, and having the predetermined model process the input text and lesson objective information in response to the objective analysis commands and output the evaluation results for the lesson to be evaluated, the objective of performing question-and-answer dialogue analysis tasks, lesson interaction analysis tasks, and curriculum standard implementation analysis tasks on the lesson without human attendance can be achieved, improving the efficiency of lesson evaluation. At the same time, since this solution is based on lesson objective information and input text converted from lesson data, it ensures objectivity and accuracy of the evaluation results. [Brief explanation of the drawing]

[0032] [Figure 1] This is a flowchart of the method for evaluating classes according to the embodiment of this application. [Figure 2] This is a flowchart of the method for evaluating classes according to the embodiment of this application. [Figure 3] This figure shows a user interface according to an embodiment of the present application. [Figure 4] This is a flowchart of the evaluation of questions asked during class according to an embodiment of this application. [Figure 5] This figure shows the structure of a lesson evaluation device according to an embodiment of this application. [Figure 6] This figure shows the structure of an electronic device according to an embodiment of this application. [Modes for carrying out the invention]

[0033] To clarify the purpose, technical solutions, and advantages of this application, specific embodiments of this application will be described in detail below, accompanied by drawings. It should be understood that the specific embodiments described herein are for the purpose of interpreting this application, but not to limit it. Also, for the sake of clarity, only some, not all, of the embodiments of this application are shown in the drawings. Before discussing the exemplary embodiments in more detail, it should be noted that some exemplary embodiments are described as processes or methods shown as flowcharts. In a flowchart, each operation (or step) is described as a sequential process, but many operations may be performed in parallel, concurrently, or simultaneously. The order of the operations can also be rearranged. The process may terminate when the operations are completed, but it may include additional steps not included in the accompanying drawings. The process may correspond to methods, functions, rules, subroutines, and subprograms.

[0034] The following describes the technical solutions in the embodiments of this application in conjunction with the drawings of the embodiments of this application, and it is clear that the embodiments described are only a part of the embodiments of this application, not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0035] The terms “first,” “second,” etc., in the specification and claims of this application are for distinguishing similar objects and are not intended to describe a specific order or sequence. The data used in this manner are interchangeable where appropriate so that the embodiments of this application may be carried out in an order other than those illustrated or described herein, and the objects distinguished as “first,” “second,” etc., are usually of the same kind and do not limit the number of objects; for example, the first object may be one or more. Also, in the specification and claims, “and / or” indicates at least one connected object, and the letter “ / ” generally indicates that related objects before and after are in an “or” relationship.

[0036] The following describes in detail the method, apparatus, device, and medium for evaluating lessons according to the embodiment of this application, through the combination of drawings, examples, and application scenarios.

[0037] Regarding the evaluation of teachers' lessons using related technologies, schools often design auditory questionnaires or scales based on relevant theories before the lesson, organize online or offline observations by teachers or instructors specializing in lesson observation, and finally, the teachers or instructors who observed the lesson summarize, discuss, and inductively analyze the teacher's interaction behavior after the lesson, based on lesson materials collected from the original lesson, and propose improvements to the lesson. However, current lesson evaluation methods have problems such as the fact that evaluations are mainly conducted by individual teachers specializing in lesson observation, which is time-consuming, and different observing teachers may have different perspectives on the same teacher's interaction behavior. Therefore, when using related technologies to evaluate lessons, there are problems with low evaluation efficiency and high subjectivity in evaluation conclusions.

[0038] In this solution, the objective of performing question-and-answer dialogue analysis tasks, lesson interaction analysis tasks, and curriculum standard implementation analysis tasks on a lesson without human attendance is achieved. This is accomplished by converting lesson data of a lesson to be evaluated into input text, receiving objective analysis commands for the lesson to be evaluated, inputting the input text and lesson objective information into a predetermined model, and having the predetermined model process the input text and lesson objective information in response to the objective analysis commands and output the evaluation results for the lesson to be evaluated. This improves the efficiency of lesson evaluation, and at the same time, since this solution is based on lesson objective information and input text converted from lesson data, it ensures the objectivity and accuracy of the evaluation results.

[0039] Figure 1 is a flowchart of a method for evaluating lessons according to an embodiment of this application. As shown in Figure 1, the method specifically includes the following steps.

[0040] In S101, lesson data and lesson objective information for the subject lesson are acquired. The lesson data includes audio data and / or video data, and the lesson objective information includes teaching objective information and curriculum standard information.

[0041] First, the usage scenarios for this solution may include analyzing and evaluating teacher question-and-answer sessions, classroom interactions, and the implementation status of curriculum standards during lessons.

[0042] Based on the above usage scenarios, it is understood that the implementer of this application may be a server-side entity that processes lesson data and evaluates teachers' lesson interaction behaviors, and that this server-side entity may include a server, terminal devices (e.g., interactive tablets, computers, smart terminals, mobile terminals, etc.).

[0043] Here, classroom interaction behaviors may include actions such as a teacher asking students questions, engaging in dialogue, or teaching the curriculum during class.

[0044] A class to be evaluated refers to a class awaiting evaluation, and may be a class awaiting evaluation of the teacher's interaction behavior during the class, or a class awaiting evaluation of the teacher's teaching content during the class. The class data for a class to be evaluated may be recorded or broadcast data of the class, or live broadcast data of the class. The class data may be the data for the entire class to be evaluated, or it may be selected class segment data for the class to be evaluated.

[0045] The lesson data includes audio data and / or video data. The audio data may include both teacher audio data and student audio data. The video data may include both teacher video data and student video data.

[0046] The aforementioned teaching objectives information includes teaching objectives information and curriculum standards information. The teaching objectives information may be the expected outcomes of educational activities, or it may be a concretization of educational objectives, teaching objectives, and curriculum objectives, or it may be the requirements and standards that teachers must meet in order to complete teaching tasks. In teaching a subject, teachers need to determine detailed teaching objectives based on curriculum objectives and teaching content in order to select teaching content and determine teaching effectiveness. Curriculum standards information may be teaching guidance documents that define the curriculum nature, curriculum objectives, content objectives, and implementation suggestions for a subject. Curriculum standards, for example, new curriculum standards for nine years of compulsory education, are described in more detail and clarity than syllabi regarding several parts such as the basic philosophy of the curriculum, curriculum objectives, and implementation suggestions for the curriculum, and they present basic learning requirements for all students.

[0047] In one embodiment, lesson data for the lesson to be evaluated can be received via a data transmission interface, and lesson objective information for the current lesson can be obtained by reading lesson data that has been previously stored or uploaded by the teacher. Here, the lesson data includes audio data and / or video data.

[0048] In one executable embodiment, the step of obtaining lesson data and lesson objective information for the lesson to be evaluated is: Steps to obtain lesson data for the course to be evaluated, The process includes the steps of querying a pre-configured database for lesson objective information that matches the lesson data, or inputting the lesson data into a pre-configured chapter matching model and causing the chapter matching model to output lesson objective information that matches the lesson data.

[0049] Here, the pre-configured database may be a course database constructed from a pre-collected educational corpus, or a database constructed from course materials uploaded by users. The pre-configured database may store, in list form, course objective information that matches the course data for each course, or course objective information that matches the course type for each course. The course type may be a type corresponding to the course content. The course type may be a type corresponding to the content of a course, such as a lecture, presentation, practice, experiment, or review. The course objective information that matches the course data may be course objective information that corresponds to at least one of the course name, course type, and subject direction from the course data. The chapter matching model may be a model for matching course data with course objective information.

[0050] In one embodiment, lesson data for the lesson to be evaluated can be obtained through a pre-configured interface. Based on at least one of the lesson name, lesson type, and subject direction from the lesson data, lesson objective information that matches the lesson data can be queried from a pre-configured database, or the lesson data can be input into a pre-configured chapter matching model so that the chapter matching model determines and outputs lesson objective information that matches the lesson data through keyword extraction, similarity matching, etc.

[0051] In the embodiments of this application, the efficiency of determining lesson objectives can be improved by querying a pre-configured database for lesson objective information that matches the lesson data, or by inputting the lesson data into a pre-configured chapter matching model and having the chapter matching model output lesson objective information that matches the lesson data.

[0052] In S102, the aforementioned lesson data is converted into input text.

[0053] Here, the input text may be text obtained by converting the aforementioned lesson data into text, teacher's question text extracted from the lesson text, or text that simultaneously includes classroom text and teacher's question text. The aforementioned lesson text may be text obtained by converting the aforementioned lesson data into text. The input text may simultaneously include the teacher's declarative sentences, the teacher's questions, the students' declarative sentences, the students' questions, etc. At the beginning of each statement text in the input text, a speaker's character prefix may be included. For example, Speaker A: "Everyone, what thoughts or feelings of the author do you think this passage expresses?", Speaker B: "It expresses the author's homesickness." It is understandable that different speakers uttering each statement text can be distinguished based on the character prefix.

[0054] In one embodiment, speech recognition technology such as ASR (Automatic Speech Recognition) can be used to convert the audio data from the lecture data into input text.

[0055] In one executable embodiment, the step of converting the lesson data into input text is: The steps include: converting the aforementioned lesson data into text to obtain the lesson text, The steps include: extracting the teacher's question text from the aforementioned lesson text, The process includes the step of using the aforementioned lesson text and the aforementioned teacher's question text as input text.

[0056] Here, the lesson text may be a text obtained by converting the aforementioned lesson data into text. The teacher's question text may be the teacher's question text in the aforementioned lesson text.

[0057] In one embodiment, the lesson data can be converted into text to obtain lesson text. Each statement text in the lesson text may begin with a speaker character prefix, and each sentence may end with punctuation. The system can identify the nouns used to refer to students within each statement text in the lesson text, determine whether the character prefix of each statement text is that of a teacher character based on the nouns used to refer to students, and further determine that a statement text in the lesson text is a teacher's question text if the character prefix is ​​that of a teacher and the punctuation at the end of the statement text is a question mark. For example, if the noun used to refer to students within a statement text is identified as "teacher," the system determines that the character prefix corresponding to that statement text is that of a student, and if the noun used to refer to students within a statement text is identified as "Mr. / Ms. ○○," the system determines that the character prefix corresponding to that statement text is that of a teacher. The lesson text and the teacher's question text are used as input text.

[0058] In the embodiments of this application, the lesson data is converted into text to obtain lesson text, the teacher's question text is extracted from the lesson text, and the lesson text and the teacher's question text are used as input text. This is advantageous for later performing specific analysis and overall analysis combining context with the teacher's questions, respectively, and improves the completeness of the input text.

[0059] In one viable embodiment, the step of extracting teacher's question text from the lesson text is: The steps include identifying the dialogue character who appears most frequently in the aforementioned lesson text, The step of determining the dialogue character with the most dialogues as the teacher character, The steps include: extracting all dialogue content of the aforementioned teacher character, This includes the step of screening the teacher's question text from all of the aforementioned dialogue content.

[0060] In one embodiment, the dialogue character with the most dialogues in the lesson text can be identified and statistically analyzed based on the frequency of occurrence of each speaker's character prefix in the lesson text. Since the lesson process mainly consists of teacher statements and teacher questions, the dialogue character with the most dialogues can be determined as the teacher character. All dialogue content of the teacher character is extracted based on the character prefix corresponding to the teacher character, and the teacher's question text in the dialogue content is screened based on punctuation at the end of each statement text in the dialogue content.

[0061] In the embodiment of this application, the dialogue character with the most frequent dialogues in the lesson text is determined as the teacher character, and the teacher's question text in the dialogue content is screened, thereby simplifying the step of extracting the teacher's question text and improving the efficiency and accuracy of the teacher's question text extraction.

[0062] In one viable embodiment, the step of screening the teacher's question text from all the dialogue content is: The steps include identifying the basic question sentences from all of the above dialogue content, The process includes the step of performing a noise reduction operation on the aforementioned basic question sentence to obtain the teacher's question text.

[0063] Here, the basic questions may be overly simple questions such as "Is that okay?" and "Is that so?", and / or questions that are repeatedly emphasized. The noise reduction operation may be an operation to delete the aforementioned basic questions.

[0064] In one embodiment, basic question sentences in all the dialogue content can be identified by searching for punctuation marks and keywords at the end of each statement text in all the dialogue content. For example, using words such as "is that okay?" and "is that so?" as keywords, statement texts with the search keywords are searched from all the dialogue content, it is identified whether the punctuation mark at the end of the searched statement text is a question mark or not, and the statement texts with a question mark at the end of the punctuation mark are determined to be basic question sentences. The basic question sentences in all the dialogue content are deleted, and statements with a period at the end of the question sentence are also deleted to obtain the teacher's question text.

[0065] In the embodiments of this application, the objective of further refining the question text is achieved by identifying basic question sentences from all of the teacher's dialogue content, performing a noise reduction operation on the basic question sentences, and obtaining the teacher's question text, thereby preventing the basic question sentences from later influencing the evaluation results of the lesson.

[0066] In S103, the objective analysis command for the evaluation subject lesson is received, and the objective analysis command includes a question-and-answer dialogue analysis command, a lesson interaction analysis command, and a curriculum standard implementation analysis command.

[0067] Here, the objective analysis command for the class under evaluation may be a command to perform a question-and-answer dialogue analysis, a classroom interaction analysis, and a curriculum standard implementation analysis on the content of the class under evaluation. The objective analysis command includes at least one of the question-and-answer dialogue analysis command, the classroom interaction analysis command, and the curriculum standard implementation analysis command. The question-and-answer dialogue analysis command is used to analyze whether the question-and-answer dialogue between the teacher and students during the class conforms to the educational objectives, the classroom interaction analysis command is used to analyze whether the classroom interaction between the teacher and students during the class conforms to the educational objectives or promotes the achievement of the educational objectives, and the curriculum standard implementation analysis command is used to analyze whether the content of the class and curriculum plan of the class under evaluation conform to the curriculum standards set forth by the Ministry of Education, Culture, Sports, Science and Technology.

[0068] In one embodiment, objective analysis commands for the subject class to be evaluated can be received via a human-computer interaction interface. For example, objective analysis commands selected or directly entered by the user for the subject class to be evaluated can be received via the human-computer interaction interface. Furthermore, depending on the curriculum type of the subject class to be evaluated, the system can read, but is not overly limited, the default objective analysis commands for that type of curriculum. Based on the objective analysis commands, the system can determine the analysis tasks for the subject class to be evaluated.

[0069] In S104, the input text and the lesson objective information are input to a predetermined model, the predetermined model processes the input text and the lesson objective information in accordance with the objective analysis command, and outputs the evaluation result of the lesson to be evaluated. The predetermined model is obtained by using the educational scene corpus as a training sample, using the question-and-answer dialogue analysis main task module, the lesson interaction analysis main task module, and the curriculum standard implementation analysis main task module as the main tasks of collaborative training, and conducting supervised fine-tuning training. The question-and-answer dialogue analysis main task module is used to respond to the question-and-answer dialogue analysis command, the lesson interaction analysis main task module is used to respond to the lesson interaction analysis command, and the curriculum standard implementation analysis main task module is used to respond to the curriculum standard implementation analysis command.

[0070] Here, the predetermined model may be a large-scale language model capable of evaluating the input text. The input limit of the predetermined model is greater than the average length of the input text, and it can perform processing corresponding to the input text based on the lesson objective information for different objective analysis commands. The evaluation result may be the result after the predetermined model has performed processing corresponding to the input text according to pre-set evaluation dimensions. The pre-set evaluation dimensions include an evaluation dimension for question effectiveness analysis, an evaluation dimension for knowledge objectives, an evaluation dimension for question avoidance, and an evaluation dimension for question optimization.

[0071] The aforementioned predetermined model is obtained by conducting supervised fine-tuning training using an educational scene corpus as a training sample, with the main task modules for question-and-answer dialogue analysis, classroom interaction analysis, and curriculum standard implementation analysis as the main tasks of collaborative training. The educational scene corpus refers to a corpus of educational resources and educational content (a corpus refers to a collection of text resources of a certain number and size), and includes lesson texts generated from lesson data actually collected offline, lesson texts generated from online lessons recorded online, texts containing educational content and question-and-answer dialogues, teachers' lesson plans, curriculum design plan texts, curriculum standard texts, teaching objectives texts, etc. Supervised fine-tuning training may also be training that uses actual labeled data to train and adjust a pre-trained language model (LLM) to better adapt it to a particular task. When supervised fine-tuning is performed, the model's weights are adjusted according to the differences from the actual labels, and this fine-tuning process allows the model to capture patterns and features specific to a particular task in the labeled data, thereby making the model more accurate and better adapted to a particular task.

[0072] The aforementioned question-and-answer dialogue analysis main task module is used in response to the question-and-answer dialogue analysis command to analyze whether the teacher's question-and-answer dialogue in the input text conforms to the educational objective information. The aforementioned classroom interaction analysis main task module is used in response to the classroom interaction analysis command to analyze whether the teacher's interaction behavior in the input text conforms to the educational objective information. The aforementioned curriculum standard implementation analysis main task module is used in response to the curriculum standard implementation analysis command to analyze whether the teaching of the curriculum in the input text is implemented in accordance with the curriculum standard information established by the Ministry of Education, Culture, Sports, Science and Technology.

[0073] In one embodiment, the predetermined model can determine an analysis task for the input text in response to the objective analysis command. The input text and the lesson objective information are input to the predetermined model so that the predetermined model processes the input text based on the lesson objective information in response to the analysis task and outputs an evaluation result for the evaluation target lesson according to a pre-set evaluation dimension.

[0074] In one executable embodiment, the question-and-answer dialogue analysis main task module includes an overall evaluation subtask module and a local evaluation subtask module, and the predetermined model, in the process of performing supervised fine-tuning training on the question-and-answer dialogue analysis main task module, processes a lesson text containing teaching objectives and teacher questions in the educational scene corpus according to the overall evaluation subtask module and the local evaluation subtask module, respectively, and outputs an evaluation result for the lesson text.

[0075] Here, the overall evaluation subtask module may be a module that focuses on the overall lesson experience, takes into account the teacher-student interaction and context, and evaluates the teacher's questions in the input text as a whole. In addition to the teacher's questions, the overall evaluation subtask module may also receive the lesson content and student responses of the lesson being evaluated, as well as the lesson text for the entire lesson. The local evaluation subtask module may be a module for evaluating the effectiveness of the teacher's current questions during the lesson. Since the local evaluation subtask module is not highly context-dependent, it may be a module that uses only the teacher's questions during the lesson as input data and can specifically evaluate the teacher's questions according to pre-set evaluation dimensions.

[0076] In one embodiment, the question-and-answer dialogue analysis main task module includes a global evaluation subtask module and a local evaluation subtask module. The predetermined model can process lesson text containing teaching objectives and teacher questions in the educational scene corpus according to the global evaluation subtask module and the local evaluation subtask module in the process of supervised fine-tuning training the question-and-answer dialogue analysis main task module, the global evaluation subtask module can combine the context of the teacher's questions in the educational scene corpus to evaluate whether the teacher's questions are consistent with the teaching objectives, and the local evaluation subtask module can evaluate whether the teacher's questions are consistent with the teaching objectives for a single question sentence of the teacher's current question and output the evaluation results for the lesson text.

[0077] In the embodiments of this application, a question-and-answer dialogue analysis main task module is provided, which includes an overall evaluation subtask module and a local evaluation subtask module. By providing supervised training to the overall evaluation subtask module and the local evaluation subtask module, respectively, based on an educational scene corpus, the objective of performing overall and specific evaluations of teachers' questions and answers can be achieved, and the accuracy of the evaluation results output by a predetermined model can be improved.

[0078] In one viable embodiment, the overall assessment subtask module includes a first overall subtask, a second overall subtask, and a third overall subtask, the first overall subtask being used in supervised fine-tuning training to overall assess the overall effectiveness of all of the teacher's questions in the lesson text; the second overall subtask being used in supervised fine-tuning training to overall assess the pre-set knowledge objectives in the lesson text; and the third overall subtask being used in supervised fine-tuning training to propose an overall optimization of the teacher's questions in the lesson text. The local assessment subtask module includes a first local subtask, a second local subtask, and a third local subtask, the first local subtask being used in supervised fine-tuning training to specifically determine the effectiveness of each of the teacher's questions in the lesson text; the second local subtask being used in supervised fine-tuning training to specifically assess the correlation between the teacher's questions and the teaching objectives in the lesson text; and the third local subtask being used in supervised fine-tuning training to optimize and rewrite the teacher's questions in the lesson text.

[0079] Here, the overall assessment subtask module includes a first overall subtask, a second overall subtask, and a third overall subtask. The first overall subtask is used in a supervised fine-tuning training process to assess the overall effectiveness of all of the teacher's questions in the lesson text. The overall assessment of overall effectiveness focuses on whether the teacher's questions can stimulate students' thinking. The second overall subtask is used in a supervised fine-tuning training process to assess the overall pre-set knowledge objectives in the lesson text. The overall assessment of the pre-set knowledge objectives focuses on the relevance of the teacher's questions to the knowledge. The third overall subtask is used in a supervised fine-tuning training process to propose an overall optimization of the teacher's questions in the lesson text. The proposal for overall optimization may refer to suggesting several corresponding optimizeable modifications by combining effectiveness and knowledge objectives.

[0080] The local assessment subtask module includes a first local subtask, a second local subtask, and a third local subtask. The first overall subtask is a classification task used in a supervised fine-tuning training process to specifically determine the effectiveness of each of the teacher's questions in the lesson text. Effectiveness may involve determining whether the teacher's current question is an open question or whether it can stimulate student thinking. The second local subtask is a classification and natural language generation task used in a supervised fine-tuning training process to specifically evaluate the degree of correlation between the teacher's questions in the lesson text and the teaching objectives. The degree of correlation requires the local assessment subtask module to determine whether the current question is related to the teaching objectives, and if so, to explain why. The third local subtask is a text rewriting task used in a supervised fine-tuning training process to optimize and rewrite the teacher's questions in the lesson text. Optimization and rewriting of the teacher's questions may involve rewriting several invalid questions into valid questions, or rewriting questions from closed questions into open questions. The above tasks all require outputting a text summary of pre-defined fields (e.g., 100-150 words), which can be done through a framework like the following:

[0081] Here, the output results of a total of six subtasks obtained by the overall evaluation subtask module and local evaluation subtask module each performing three corresponding subtasks can be integrated and inductively analyzed according to four evaluation dimensions: question effectiveness analysis, knowledge objectives, question avoidance, and question optimization, and further output according to the four evaluation dimensions. Here, question effectiveness analysis may be used to evaluate the effectiveness of teachers' questions in the current curriculum, i.e., whether they can stimulate students' thinking; knowledge objectives may be used to evaluate the relationship between teachers' questions and teaching objectives in the current curriculum; question avoidance may be used to find avoidable questions for teachers in the current curriculum; and question optimization may be used to find directions for optimizing teachers' questions in the current curriculum.

[0082] Here, the evaluation dimension of the question effectiveness analysis may be a combination of two parts: an overall analysis of the effectiveness of the questions, which is evaluated overall, and a judgment of effectiveness, which is evaluated specifically. In one example, the framework may take the following form: TIFF2026512185000002.tif36164

[0083] The evaluation dimension for knowledge objectives may be a combination of two parts: an overall analysis of knowledge objectives, which are evaluated holistically, and the correlation with specific teaching objectives, which are evaluated specifically. In one example, the framework may take the following form: TIFF2026512185000003.tif36164

[0084] Regarding the evaluation dimensions for avoidable questions, inappropriate questions during class can be cited. In this application, questions to be avoided are defined as those that are "unrelated to the educational content" and "fail to stimulate students' thinking." The judgment logic is that, as an effective judgment in specific analysis, the final result will be questions that are judged to have low effectiveness and low correlation with teaching.

[0085] The evaluation dimension for question optimization needs to include optimizations to the current question, such as the question rewriting portion in the specific evaluation and the question optimization proposal portion in the overall evaluation. The final result will be output in a modification-proposal structure, and in one example, the framework may take the following form. TIFF2026512185000004.tif30164

[0086] Finally, the four evaluation dimensions mentioned above are analyzed and summarized in an analysis report, which is then output as the final result.

[0087] In the embodiments of this application, a whole evaluation subtask module containing three whole subtasks and a local evaluation subtask module containing three local subtasks are provided, thereby jointly conducting supervised training for a total of six subtasks (three whole subtasks and three local subtasks). In the training process, the first whole subtask and the first local subtask collaborate to produce evaluation results in the dimension of question effectiveness analysis, the second whole subtask and the second local subtask collaborate to produce evaluation results in the dimension of knowledge objectives, and the third whole subtask The first and third local subtasks collaborate to output evaluation results in the dimension of question optimization, while the first and second local subtasks collaborate to output evaluation results in the dimension of question avoidance. In other words, all subtasks performed collaboratively in the overall evaluation subtask module and the local evaluation subtask module can output evaluation results of teachers' questions in the evaluated lesson according to four evaluation dimensions: question effectiveness analysis, knowledge objectives, question avoidance, and question optimization. This ensures objectivity in lesson evaluation and further improves the comprehensiveness of the analysis of teachers' questions during lessons.

[0088] The technical solution provided by the embodiment of this application acquires lesson data and lesson objective information for a lesson to be evaluated, the lesson data includes audio data and / or video data, the lesson objective information includes teaching objective information and curriculum standard information, the lesson data is converted into input text, an objective analysis command for the lesson to be evaluated is received, the objective analysis command includes a question-and-answer dialogue analysis command, a lesson interaction analysis command and a curriculum standard implementation analysis command, the input text and the lesson objective information are input into a predetermined model, the predetermined model processes the input text and the lesson objective information in response to the objective analysis command, and outputs the evaluation result for the lesson to be evaluated. The aforementioned predetermined model is obtained by using an educational scene corpus as a training sample, and by conducting supervised fine-tuning training using the following modules as the main tasks of collaborative training: the Q&A dialogue analysis main task module, the classroom interaction analysis main task module, and the curriculum standard implementation analysis main task module. The Q&A dialogue analysis main task module is used to respond to the Q&A dialogue analysis command, the classroom interaction analysis main task module is used to respond to the classroom interaction analysis command, and the curriculum standard implementation analysis main task module is used to respond to the curriculum standard implementation analysis command.The above method for evaluating classes solves the problems of low efficiency and high subjectivity in evaluation conclusions when evaluating class interaction behavior using related technologies. By converting the class data of the class to be evaluated into input text, receiving objective analysis commands for the class to be evaluated, inputting the input text and the class objective information into a predetermined model, and having the predetermined model process the input text and the class objective information in accordance with the objective analysis commands and outputting the evaluation results for the class to be evaluated, the objective of performing question-and-answer dialogue analysis tasks, class interaction analysis tasks, and curriculum standard implementation analysis tasks on the class without artificial listening can be achieved, improving the efficiency of class evaluation, and at the same time, since this solution is based on class evaluations conducted on input text converted from class objective information and class data, objectivity and accuracy of the evaluation results can be ensured.

[0089] Figure 2 is a flowchart of a method for evaluating lessons according to an embodiment of this application. As shown in Figure 2, the method specifically includes the following steps.

[0090] In S201, lesson data and lesson objective information for the subject lesson are acquired. The lesson data includes audio data and / or video data, and the lesson objective information includes teaching objective information and curriculum standard information.

[0091] In S202, the aforementioned lesson data is converted into input text.

[0092] In S203, the objective analysis command for the subject class to be evaluated is received, and the objective analysis command includes a question-and-answer dialogue analysis command, a class interaction analysis command, and a curriculum standard implementation analysis command.

[0093] In S2041, if it is confirmed that the received objective analysis command is a question-and-answer dialogue analysis command, the input text and the teaching objective information are input into a predetermined model, the predetermined model responds to the question-and-answer dialogue analysis command and executes the question-and-answer dialogue analysis main task module based on the input text and the teaching objective information, and outputs the evaluation results of the evaluation target lesson in the question-and-answer dialogue analysis main task module.

[0094] In one embodiment, if it is confirmed that the received objective analysis command is a question-and-answer dialogue analysis command, the input text and the teaching objective information are input into a predetermined model, and the predetermined model responds to the question-and-answer dialogue analysis command and executes the question-and-answer dialogue analysis main task module. The question-and-answer dialogue analysis main task module processes the input text based on the teaching objective information, analyzes whether the teacher's question-and-answer dialogue in the input text conforms to the teaching objective information, and further outputs the evaluation result of the evaluation target lesson in the question-and-answer dialogue analysis main task module.

[0095] In one executable embodiment, if it is confirmed that the received objective analysis command is a question-and-answer dialogue analysis command, the input target text and the teaching objective information are input into a predetermined model, the predetermined model responds to the question-and-answer dialogue analysis command, executes the question-and-answer dialogue analysis main task module based on the input target text and the teaching objective information, and outputs the evaluation result of the evaluation target lesson in the question-and-answer dialogue analysis main task module, If it is confirmed that the received objective analysis command is a question-and-answer dialogue analysis command, the predetermined model is confirmed to respond to the question-and-answer dialogue analysis command, and the predetermined model is set to wait for the execution of objective evaluation tasks in each type of evaluation dimension in the question-and-answer dialogue analysis main task module. The steps to determine the objective evaluation task, The main task module for question-and-answer dialogue analysis inputs the input text and the teaching objective information into a predetermined model, and includes the step of executing the objective evaluation task based on the input text and the teaching objective information, and outputting the evaluation results of the evaluation target class.

[0096] Here, the evaluation dimensions may be four: question effectiveness analysis, knowledge-based goals, question avoidance, and question optimization. The goal evaluation task may be a task that corresponds to at least one of the four evaluation dimensions.

[0097] In one embodiment, if it is confirmed that the received objective analysis command is a question-and-answer dialogue analysis command, the model can determine whether to respond to the question-and-answer dialogue analysis command based on whether it has received a response signal fed back by the predetermined model. If the predetermined model confirms that it will respond to the question-and-answer dialogue analysis command, the predetermined model enters a state of waiting to execute objective evaluation tasks in each type of evaluation dimension in the question-and-answer dialogue analysis main task module. An evaluation command can be received by the user via a human-computer interaction interface, and an objective evaluation task can be determined in accordance with the evaluation command. The input text and the teaching objective information are input to the predetermined model, the question-and-answer dialogue analysis main task module executes the objective evaluation task based on the input text and the teaching objective information, and outputs the evaluation result of the evaluation target lesson.

[0098] Figure 3 shows a user interface according to an embodiment of the present application. As shown in Figure 3, the left side of the interface includes three main tasks, respectively: question-and-answer dialogue, classroom interaction, and implementation of new curriculum standards. When the user selects "Question-and-Answer Dialogue" on the left side of the interface, the right side of the interface may display evaluation results, with the default evaluation dimension being "Question Effectiveness Analysis." Options below the evaluation results displayed on the right side include which questions fit the knowledge objectives, which questions are inappropriate, optimizing the questions, and viewing all questions. The user can select the appropriate option according to their viewing needs. Upon receiving the user's selection of the "Optimize Questions" option, it can be determined that the objective evaluation task of the predetermined model is question optimization, and the input text and the teaching objective information are input into the predetermined model, thereby allowing the question-and-answer dialogue analysis main task module to perform the question optimization task based on the input text and the teaching objective information and output optimization suggestions for the evaluated lesson viewed by the user.

[0099] In this solution, after confirming that a question-and-answer dialogue analysis command has been received and that a predetermined model will respond to the command, the predetermined model enters a state where it waits for the execution of the question-and-answer dialogue analysis main task module. In this case, the predetermined model determines the target evaluation tasks in various types of evaluation dimensions within the question-and-answer dialogue analysis main task module, thereby executing the target evaluation tasks based on the input text and the teaching objective information. This achieves the effect of analyzing the teacher's questions during lessons based on different evaluation dimensions.

[0100] Figure 4 is a flowchart of the evaluation of questions during class according to an embodiment of this application. As shown in Figure 4, the data processing module performs ASR transformation on the actual class data to obtain the class text. The class text and teaching objectives are input into a predetermined model, which can perform a comprehensive evaluation of the teacher's questions in the class text, combining context, based on three dimensions: overall question effectiveness evaluation, overall knowledge objective evaluation, and question optimization suggestions. In addition, speaker identification is performed on the class text, the teacher's dialogue is extracted, and question sentences are extracted and / or filtered from the teacher's dialogue to obtain effective question sentences for the teacher's questions. Effective question sentences and teaching objectives are input into the predetermined model so that the predetermined model performs a specific evaluation of a single question sentence for the teacher's questions in the class text based on three dimensions: effectiveness judgment, question-teaching correlation, and question rewriting.

[0101] As shown in Figure 4, the main task module for question-and-answer dialogue analysis includes an overall evaluation subtask module and a local evaluation subtask module. The aforementioned step of determining the objective evaluation task is, A step of obtaining a first type of evaluation prompt command, the first type of evaluation prompt command is used to prompt a predetermined model to output an evaluation result of the lesson under evaluation according to an evaluation dimension of question effectiveness analysis, the evaluation dimension of question effectiveness analysis includes an evaluation of the effectiveness of the teacher's questions in the lesson under evaluation, A step in which, in response to the first type of evaluation prompt command, it is determined that a first overall subtask in the overall evaluation subtask module and a first local subtask in the local evaluation subtask module jointly constitute a target evaluation task, wherein the first overall subtask is used to overall evaluate the overall effectiveness of all of the teacher's questions in the lesson being evaluated, and the first local subtask is used to specifically determine the effectiveness of each of the teacher's questions in the lesson being evaluated. and / or, A step of obtaining a second type of evaluation prompt command, the second type of evaluation prompt command being used to prompt a predetermined model to output an evaluation result of the lesson being evaluated according to an evaluation dimension of knowledge objectives, the evaluation dimension of knowledge objectives including an evaluation of the relevance between the teacher's questions and the teaching objectives in the lesson being evaluated, A step in which, in response to the second type of evaluation prompt command, it is determined that the second overall subtask in the overall evaluation subtask module and the second local subtask in the local evaluation subtask module jointly constitute an objective evaluation task, wherein the second overall subtask is used to evaluate the overall pre-set knowledge objectives in the lesson being evaluated, and the second local subtask is used to specifically evaluate the degree of correlation between the teacher's questions and the teaching objectives in the lesson being evaluated. and / or, A step of obtaining a third type of evaluation prompt command as the target evaluation task, wherein the third type of evaluation prompt command is used to prompt a predetermined model to output an evaluation result of the lesson to be evaluated according to an evaluation dimension of question avoidance, the evaluation dimension of question avoidance includes a step of identifying teacher avoidable questions in the lesson to be evaluated, A step in which, in response to the third type of evaluation prompt command, it is determined that a first local subtask and a second local subtask in the local evaluation subtask module jointly constitute an objective evaluation task, wherein the first local subtask is used to specifically determine the effectiveness of each question asked by the teacher in the lesson being evaluated, and the second local subtask is used to specifically evaluate the degree of correlation between the teacher's questions and the teaching objectives in the lesson being evaluated. and / or, A step of obtaining a fourth type of evaluation prompt command as the target evaluation task, wherein the fourth type of evaluation prompt command is used to prompt a predetermined model to output an evaluation result of the lesson to be evaluated according to an evaluation dimension of question optimization, the evaluation dimension of question optimization includes finding teacher-optimizable questions in the lesson to be evaluated, In response to the fourth type of evaluation prompt command, the step of determining that a third global subtask in the global evaluation subtask module and a third local subtask in the local evaluation subtask module jointly constitute a target evaluation task, wherein the third global subtask is used to propose an overall optimization of the teacher's questions in the lesson being evaluated, and the third local subtask is used to optimize and rewrite the teacher's questions in the lesson being evaluated.

[0102] In the first embodiment, a first type of evaluation prompt command can be obtained via a user interface. The first type of evaluation prompt command is used to prompt a predetermined model to output an evaluation result of the lesson under evaluation according to the evaluation dimension of the question effectiveness analysis. The evaluation dimension of the question effectiveness analysis includes an evaluation of the effectiveness of the teacher's questions in the lesson under evaluation. In response to the first type of evaluation prompt command, it is determined that the first overall subtask in the overall evaluation subtask module and the first local subtask in the local evaluation subtask module jointly constitute the target evaluation task. As indicated by arrows "101" and "201" in Figure 4, "Overall Analysis of Question Effectiveness" in the figure represents the first overall subtask. The first overall subtask is used to give an overall evaluation of the overall effectiveness of all of the teacher's questions in the lesson under evaluation, and the first local subtask is used to specifically determine the effectiveness of each of the teacher's questions in the lesson under evaluation. "Effectiveness Determination" in the figure represents the first local subtask. The first local subtask is used to determine whether each of the teacher's questions can stimulate student thinking. In the above objective evaluation task, the overall evaluation result and local evaluation result of the evaluated class that are output can be combined as the first type of evaluation result. The first type of evaluation result is the evaluation result of the evaluated class output by the predetermined model according to the evaluation dimension of the question effectiveness analysis, and may be output according to the output framework of the question effectiveness analysis dimension, which will not be explained here.

[0103] In the second embodiment, a second type of evaluation prompt command can be obtained via a user interface. The second type of evaluation prompt command is used to prompt the predetermined model to output an evaluation result of the lesson under evaluation according to an evaluation dimension of knowledge objectives. The evaluation dimension of knowledge objectives includes an evaluation of the relevance between the teacher's questions and the teaching objectives in the lesson under evaluation. In response to the second type of evaluation prompt command, it is determined that the second overall subtask in the overall evaluation subtask module and the second local subtask in the local evaluation subtask module jointly constitute the objective evaluation task. As indicated by arrows "102" and "202" in Figure 4, "Overall Analysis of Knowledge Objectives" in the figure represents the second overall subtask. The second overall subtask is used to evaluate the pre-set knowledge objectives in the lesson under evaluation as a whole. "Correlation of Teaching Objectives" in the figure represents the second local subtask. The second local subtask is used to specifically evaluate the correlation between the teacher's questions and the teaching objectives in the lesson under evaluation. In the objective evaluation task described above, the overall evaluation result and local evaluation result of the evaluated class that are output can be combined to produce a second type of evaluation result. The second type of evaluation result is the evaluation result of the evaluated class output by the predetermined model according to the evaluation dimension of the knowledge objective, and may be output according to the output framework of the knowledge objective analysis dimension described above, but the explanation is omitted here.

[0104] In the third embodiment, a third type of evaluation prompt command can be obtained as the target evaluation task via a user interface. The third type of evaluation prompt command is used to prompt the predetermined model to output evaluation results for the lesson under evaluation according to the evaluation dimension of question avoidance. The evaluation dimension of question avoidance includes identifying avoidable questions by the teacher in the lesson under evaluation. In this embodiment, avoidable questions are defined as questions that are "unrelated to the educational content" and "cannot stimulate student thinking," and in one embodiment, the judgment logic uses questions that are judged to have low effectiveness and low correlation with education as the final question avoidance result, as an effectiveness judgment in the analysis results of the local subtask. In response to the third type of evaluation prompt command, it is decided that the first local subtask and the second local subtask in the local evaluation subtask module jointly constitute the target evaluation task, the first local subtask is used to specifically determine the effectiveness of each of the teacher's questions in the lesson under evaluation, and the second local subtask is used to specifically evaluate the correlation between the teacher's questions and the educational objectives in the lesson under evaluation.

[0105] In the fourth embodiment, a fourth type of evaluation prompt command can be obtained as the target evaluation task via a user interface. The fourth type of evaluation prompt command is used to prompt the predetermined model to output evaluation results for the lesson under evaluation according to the evaluation dimension of question optimization. The evaluation dimension of question optimization includes finding teacher questions that can be optimized in the lesson under evaluation and the optimization results of the current teacher questions. In response to the fourth type of evaluation prompt command, it is determined that the third global subtask in the global evaluation subtask module and the third local subtask in the local evaluation subtask module jointly constitute the target evaluation task. As indicated by arrows "103" and "203" in Figure 4, "Propose Question Optimization" in the figure represents the third global subtask. The third global subtask is used to propose an overall optimization of the teacher's questions in the lesson under evaluation. "Rewrite Questions" in the figure represents the third local subtask. The third global subtask is used to optimize and rewrite the teacher's questions in the lesson under evaluation. In the above objective evaluation task, the overall evaluation result and local evaluation result of the evaluated lesson can be combined to form a third type of evaluation result. The third type of evaluation result is the optimization result of the evaluated lesson output by the predetermined model according to the evaluation dimension of question optimization, and may be output according to the output framework of the question optimization dimension, which will not be explained here.

[0106] It is understandable that this application allows for the evaluation of a target lesson through the above embodiments according to at least one dimension, and that the target lesson can also be evaluated simultaneously according to the above four dimensions. In other words, this application provides 15 solutions for evaluating a target lesson based on different dimensions.

[0107] In the embodiments of this application, by simultaneously or selectively evaluating the content of the question-and-answer dialogue in the lessons being evaluated based on different evaluation dimensions and different evaluation ranges, it is possible to evaluate the teacher's questions in the lessons being evaluated in different dimensions, assuming the use of a unified model. This avoids the problem in related technologies where training results in different dimensions become isolated from each other, as each evaluation dimension requires separate training using lesson data. This can more effectively reduce the amount of iterative model training work, increase the relevance between training tasks in different evaluation dimensions, and further improve the objectivity of lesson evaluation.

[0108] In S2042, if it is confirmed that the received objective analysis command is a lesson interaction analysis command, the input text and the teaching objective information are input into a predetermined model, the predetermined model responds to the lesson interaction analysis command and executes the lesson interaction analysis main task module based on the input text and the teaching objective information, and outputs the evaluation results of the lesson to be evaluated in the lesson interaction analysis main task module.

[0109] In one embodiment, when it is confirmed that the received objective analysis command is a lesson interaction analysis command, the input text and the teaching objective information are input into a predetermined model, and the predetermined model can respond to the lesson interaction analysis command. The predetermined model can execute the lesson interaction analysis main task module based on the input text and the teaching objective information. The lesson interaction analysis main task module responds to the lesson interaction analysis command by analyzing whether the teacher's interaction behavior in the input text conforms to the teaching objective information, and further outputs the evaluation result of the lesson to be evaluated in the lesson interaction analysis main task module.

[0110] In S2043, if it is confirmed that the received objective analysis command is a curriculum standard implementation analysis command, the input target text and the curriculum standard information are input into a predetermined model, the predetermined model responds to the curriculum standard implementation analysis command and executes the curriculum standard implementation analysis main task module based on the input target text and the curriculum standard information, and outputs the evaluation results of the evaluation target class in the curriculum standard implementation analysis main task module.

[0111] In one embodiment, when it is confirmed that the received objective analysis command is a curriculum standard implementation analysis command, the input target text and the curriculum standard information are input into a predetermined model, and the predetermined model responds to the curriculum standard implementation analysis command. The curriculum standard implementation analysis main task module responds to the curriculum standard implementation analysis command and analyzes whether the teaching of the curriculum in the input target text is implemented in accordance with the curriculum standard information established by the Ministry of Education, Culture, Sports, Science and Technology, and further outputs the evaluation result of the evaluation target lesson in the curriculum standard implementation analysis main task module.

[0112] The technical solution provided by the embodiment of this application inputs target text and lesson objective information into a predetermined model in response to different types of objective analysis commands received. The predetermined model responds to different objective analysis commands by executing a corresponding analysis main task module based on curriculum standard information or teaching objective information within the lesson objective information and the input text. The evaluation results of the lesson to be evaluated are output in the correspondingly selected main task module. This achieves the objective of performing evaluations corresponding to the lesson to be evaluated according to the evaluation tasks in different types of main task modules, thereby improving the objectivity of the evaluation of lesson interaction behavior and the user experience when viewing the evaluation results.

[0113] Figure 5 shows the structure of a lesson evaluation device according to an embodiment of this application. As shown in Figure 5, the lesson evaluation device specifically comprises: Used to acquire lesson data and lesson objective information for the lessons to be evaluated, the lesson data includes audio data and / or video data, and the lesson objective information includes teaching objective information and curriculum standard information, and a data acquisition module 501 is used. A data conversion module 502 for converting the aforementioned lesson data into input text, Used to receive objective analysis commands for the aforementioned evaluation subject class, the objective analysis commands include a command receiving module 503 which includes a question-and-answer dialogue analysis command, a class interaction analysis command, and a curriculum standard implementation analysis command. The input target text and the lesson objective information are input into a predetermined model, the predetermined model processes the input target text and the lesson objective information in accordance with the objective analysis command, and is used to output the evaluation results of the lesson to be evaluated. The predetermined model is obtained by using an educational scene corpus as a training sample, using a question-and-answer dialogue analysis main task module, a lesson interaction analysis main task module, and a curriculum standard implementation analysis main task module as the main tasks of collaborative training, and conducting supervised fine-tuning training. The question-and-answer dialogue analysis main task module is used to respond to the question-and-answer dialogue analysis command, the lesson interaction analysis main task module is used to respond to the lesson interaction analysis command, and the curriculum standard implementation analysis main task module is used to respond to the curriculum standard implementation analysis command.

[0114] In one embodiment, the evaluation module 504 is A question-and-answer dialogue analysis unit is provided to confirm that the received objective analysis command is a question-and-answer dialogue analysis command, input the input target text and the teaching objective information into a predetermined model, have the predetermined model respond to the question-and-answer dialogue analysis command, execute the question-and-answer dialogue analysis main task module based on the input target text and the teaching objective information, and output the evaluation results of the evaluation target lesson in the question-and-answer dialogue analysis main task module. A class interaction analysis unit that, upon confirming that the received objective analysis command is a class interaction analysis command, inputs the input target text and the teaching objective information into a predetermined model, the predetermined model responds to the class interaction analysis command, executes the class interaction analysis main task module based on the input target text and the teaching objective information, and outputs the evaluation results of the class to be evaluated in the class interaction analysis main task module, The system includes a curriculum standard implementation analysis unit that, upon confirming that the received objective analysis command is a curriculum standard implementation analysis command, inputs the input target text and the curriculum standard information into a predetermined model, the predetermined model responds to the curriculum standard implementation analysis command, executes the curriculum standard implementation analysis main task module based on the input target text and the curriculum standard information, and outputs the evaluation results of the evaluation target class in the curriculum standard implementation analysis main task module.

[0115] In one embodiment, the question-and-answer dialogue analysis unit is specifically: If it is confirmed that the received objective analysis command is a question-and-answer dialogue analysis command, the predetermined model is confirmed to respond to the question-and-answer dialogue analysis command, and the predetermined model is set to wait for the execution of objective evaluation tasks in each type of evaluation dimension in the question-and-answer dialogue analysis main task module. Determine the target evaluation task, The input text and the teaching objective information are input into a predetermined model, and the question-and-answer dialogue analysis main task module is used to perform the objective evaluation task based on the input text and the teaching objective information, and to output the evaluation results of the evaluation target class.

[0116] In one embodiment, the main task module for question-and-answer dialogue analysis includes an overall evaluation subtask module and a local evaluation subtask module. The aforementioned question-and-answer dialogue analysis unit is specifically used to perform the following operations: A first type of evaluation prompt command is obtained, and the first type of evaluation prompt command is used to prompt the predetermined model to output an evaluation result of the lesson under evaluation according to the evaluation dimension of the question effectiveness analysis, the evaluation dimension of the question effectiveness analysis includes an evaluation of the effectiveness of the teacher's questions in the lesson under evaluation, In response to the first type of evaluation prompt command, the first overall subtask in the overall evaluation subtask module and the first local subtask in the local evaluation subtask module decide to jointly constitute a target evaluation task, the first overall subtask is used to overall evaluate the overall effectiveness of all of the teacher's questions in the evaluation lesson, and the first local subtask is used to specifically determine the effectiveness of each of the teacher's questions in the evaluation lesson. and / or, A second type of evaluation prompt command is obtained, which is used to prompt the predetermined model to output an evaluation result of the lesson being evaluated according to an evaluation dimension of knowledge objectives, the evaluation dimension of knowledge objectives includes an evaluation of the relevance between the teacher's questions and the teaching objectives in the lesson being evaluated. In response to the second type of evaluation prompt command, the second overall subtask in the overall evaluation subtask module and the second local subtask in the local evaluation subtask module decide to jointly constitute an objective evaluation task, the second overall subtask is used to evaluate the pre-set knowledge objectives in the lesson being evaluated as a whole, and the second local subtask is used to evaluate the degree of correlation between the teacher's questions and the teaching objectives in the lesson being evaluated. and / or, A third type of evaluation prompt command is acquired as the target evaluation task, and the third type of evaluation prompt command is used to prompt the predetermined model to output an evaluation result of the lesson being evaluated according to the question avoidance evaluation dimension, the question avoidance evaluation dimension includes identifying avoidable questions for the teacher in the lesson being evaluated. In response to the third type of evaluation prompt command, the first local subtask and the second local subtask in the local evaluation subtask module decide to jointly constitute an objective evaluation task, the first local subtask is used to specifically determine the effectiveness of each of the teacher's questions in the lesson being evaluated, and the second local subtask is used to specifically evaluate the degree of correlation between the teacher's questions and the teaching objectives in the lesson being evaluated. and / or, A fourth type of evaluation prompt command is acquired as the target evaluation task, and the fourth type of evaluation prompt command is used to prompt the predetermined model to output the evaluation results of the lesson under evaluation according to the evaluation dimension of question optimization, the evaluation dimension of question optimization includes finding teacher-optimizable questions in the lesson under evaluation. In response to the fourth type of evaluation prompt command, the third global subtask in the global evaluation subtask module and the third local subtask in the local evaluation subtask module decide to jointly constitute a target evaluation task, the third global subtask is used to propose an overall optimization of the teacher's questions in the lesson being evaluated, and the third local subtask is used to optimize and rewrite the teacher's questions in the lesson being evaluated.

[0117] In one embodiment, the main task module for question-and-answer dialogue analysis includes an overall evaluation subtask module and a local evaluation subtask module, and the predetermined model, in the process of performing supervised fine-tuning training on the main task module for question-and-answer dialogue analysis, processes the lesson text containing the teaching objectives and teacher questions in the educational scene corpus according to the overall evaluation subtask module and the local evaluation subtask module, respectively, and outputs an evaluation result for the lesson text.

[0118] In one embodiment, the overall assessment subtask module includes a first overall subtask, a second overall subtask, and a third overall subtask, the first overall subtask being used in supervised fine-tuning training to overall assess the overall effectiveness of all of the teacher's questions in the lesson text; the second overall subtask being used in supervised fine-tuning training to overall assess the pre-set knowledge objectives in the lesson text; and the third overall subtask being used in supervised fine-tuning training to propose an overall optimization of the teacher's questions in the lesson text. The local assessment subtask module includes a first local subtask, a second local subtask, and a third local subtask, the first local subtask being used in supervised fine-tuning training to specifically determine the effectiveness of each of the teacher's questions in the lesson text; the second local subtask being used in supervised fine-tuning training to specifically assess the correlation between the teacher's questions and the teaching objectives in the lesson text; and the third local subtask being used in supervised fine-tuning training to optimize and rewrite the teacher's questions in the lesson text.

[0119] In one embodiment, the data acquisition module 501 specifically is: We obtained the lesson data for the course to be evaluated. It is used to query a pre-configured database for lesson objective information that matches the lesson data, or to input the lesson data into a pre-configured chapter matching model and have the chapter matching model output lesson objective information that matches the lesson data.

[0120] In one embodiment, the data conversion module 502 specifically is: The aforementioned lesson data is converted into text to obtain the lesson text, From the aforementioned lesson text, extract the teacher's question text. The aforementioned lesson text and the aforementioned teacher's question text are used as input text.

[0121] In one embodiment, the data conversion module 502 specifically is: Identify the dialogue character with the most frequent dialogues in the aforementioned course text, The character with the most dialogues is selected as the teacher character. Extract all dialogue content of the aforementioned teacher character, This is used to screen the teacher's question text from all of the aforementioned dialogue content.

[0122] In one embodiment, the data conversion module 502 specifically is: Identify the basic question sentences from all of the above dialogue content, This process is used to perform noise reduction on the aforementioned basic question sentences in order to obtain the teacher's question text.

[0123] In the technical solution provided by the embodiments of this application, a data acquisition module is used to acquire lesson data and lesson objective information for a lesson to be evaluated, the lesson data includes audio data and / or video data, the lesson objective information includes teaching objective information and curriculum standard information, a data transformation module is used to convert the lesson data into input text, a command receiving module is used to receive objective analysis commands for the lesson to be evaluated, the objective analysis commands include question-and-answer dialogue analysis commands, lesson interaction analysis commands and curriculum standard implementation analysis commands, and an evaluation module inputs the input text and the lesson objective information into a predetermined model, and the predetermined model processes the input text and previous information in response to the objective analysis commands. The specified model is used to process the lesson objective information and output the evaluation results of the lessons to be evaluated. The specified model is obtained by using the educational scene corpus as a training sample, and by using the question-and-answer dialogue analysis main task module, the lesson interaction analysis main task module, and the curriculum standard implementation analysis main task module as the main tasks of collaborative training, and by conducting supervised fine-tuning training. The question-and-answer dialogue analysis main task module is used to respond to the question-and-answer dialogue analysis command, the lesson interaction analysis main task module is used to respond to the lesson interaction analysis command, and the curriculum standard implementation analysis main task module is used to respond to the curriculum standard implementation analysis command.The above-described lesson evaluation device solves the problem of low efficiency and high subjectivity in evaluation conclusions when evaluating lesson interaction behavior using related technologies. By converting lesson data of the lesson to be evaluated into input text, receiving objective analysis commands for the lesson to be evaluated, inputting the input text and lesson objective information into a predetermined model, and having the predetermined model process the input text and lesson objective information in accordance with the objective analysis commands, and outputting the evaluation results for the lesson to be evaluated, the objective of performing question-and-answer dialogue analysis tasks, lesson interaction analysis tasks, and curriculum standard implementation analysis tasks for the lesson can be achieved without human attendance, thereby improving the efficiency of lesson evaluation, and at the same time, since this solution is based on lesson objective information and input text for lesson evaluation, objectivity and accuracy of the evaluation results can be ensured.

[0124] The teaching evaluation device in the embodiments of this application may be a device, a component within a terminal, an integrated circuit, or a chip. The device may be a mobile electronic device or a non-mobile electronic device. For example, a mobile electronic device may be a mobile phone, tablet computer, notebook computer, Palm computer, in-vehicle electronic device, wearable device, ultra-mobile personal computer (UMPC), netbook, or personal digital assistant (PDA), and a non-mobile electronic device may be a server, network attached storage (NAS), personal computer (PC), television, ATM, or self-service machine, and the embodiments of this application are not specifically limited.

[0125] The teaching evaluation device in the embodiments of this application may be a device equipped with an operating system. The operating system may be an (Android) operating system, an iOS operating system, or any other possible operating system, and is not specifically limited to the embodiments of this application.

[0126] The teaching evaluation device according to the embodiment of this application can implement each process realized by the embodiment of the above method, and to avoid duplication, repeated explanations are omitted here.

[0127] Figure 6 shows the structure of an electronic device according to an embodiment of the present application. As shown in Figure 6, the embodiment of the present application further provides an electronic device 600 comprising a processor 601, a memory 602, and a program or instruction stored in the memory 602 and executable by the processor 601. When the program or instruction is executed by the processor 601, each process of the embodiment of the above-described lesson evaluation method is realized and the same technical effect can be achieved. To avoid duplication, repeated explanations are omitted here.

[0128] For the purposes of this application, the electronic devices in the embodiments include the mobile electronic devices and non-mobile electronic devices described above.

[0129] The embodiments of this application further provide a readable storage medium in which a program or instruction is stored, and when the program or instruction is executed by a processor, each process of the embodiments of the above-described lesson evaluation method is realized and the same technical effects can be achieved. To avoid duplication, repeated explanations are omitted here.

[0130] Here, the processor is the processor within the electronic device in the above embodiment. The readable storage medium includes computer-readable storage media such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0131] Embodiments of this application further provide a chip comprising a processor and a communication interface, the communication interface being coupled to the processor, the processor executing a program or instructions and used to implement each process of the embodiment of the above-described method for evaluating instruction, and achieving the same technical effects, and to avoid duplication, repeated explanations are omitted here.

[0132] It should be understood that the chips referred to in the embodiments of this application may also be called system-level chips, system chips, chip systems, or system-on-a-chip.

[0133] It should be noted that, in this specification, the terms “compose,” “incorporate,” or any variation thereof are intended to cover non-exclusive inclusion, so a process, method, article, or apparatus containing a set of elements includes not only such elements but also other elements not explicitly shown, or further elements specific to such process, method, article, or apparatus. Unless otherwise specified, an element limited by the phrase “one…composes” does not preclude the existence of other identical elements in a process, method, article, or apparatus containing that element. Furthermore, it should be noted that the scope of methods and apparatus in embodiments of this application is not limited to performing functions in the order illustrated or discussed, but may include performing functions essentially simultaneously or in reverse order, depending on the function, for example, performing the described method in an order different from the order of the described method, and various steps may be added, omitted, or combined. Also, features described with reference to some examples may be combined in other examples.

[0134] Through the above description of embodiments, those skilled in the art will clearly understand that the methods according to the embodiments may be implemented by combining software and the necessary hardware platform, or of course, by hardware, and that in many cases the former is a better embodiment. Based on this understanding, the technical solutions of the present application may be embodied substantially or in part in the form of a computer software product, which is stored on a storage medium (e.g., ROM / RAM, magnetic disk, optical disk) containing several commands for causing a terminal (which may be a mobile phone, computer, server, or network device, etc.) to perform the methods according to each embodiment of the present application.

[0135] Although embodiments of this application have been described above in combination with the attached drawings, this application is not limited to the above-described specific embodiments. The above-described specific embodiments are merely illustrative and not restrictive. Those skilled in the art may make many forms based on the teachings of this application without departing from the gist of this application and the scope of protection of the claims, and all of these fall within the scope of protection of this application.

[0136] The above are merely preferred embodiments and applicable technical principles of this application. This application is not limited to the specific embodiments described herein, and those skilled in the art will understand that various obvious modifications, readjustments, and substitutions can be made without departing from the scope of protection of this application. Accordingly, although this application is described in more detail through the above embodiments, this application is not limited to the above embodiments and may include many other equivalent embodiments without departing from the concept of this application, and the scope of this application is determined by the appended claims.

Claims

1. A method of evaluating a course, A step of acquiring lesson data and lesson objective information for a lesson to be evaluated, wherein the lesson data includes audio data and / or video data, and the lesson objective information includes teaching objective information and curriculum standard information. The steps include converting the aforementioned lesson data into input text, A step of receiving an objective analysis command for the subject class to be evaluated, wherein the objective analysis command includes a question-and-answer dialogue analysis command, a class interaction analysis command, and a curriculum standard implementation analysis command, A method for evaluating a lesson, comprising the steps of: inputting the input target text and the lesson objective information into a predetermined model; the predetermined model processing the input target text and the lesson objective information in accordance with the objective analysis command; and outputting the evaluation result of the lesson to be evaluated, wherein the predetermined model is obtained by using an educational scene corpus as a training sample, using a question-and-answer dialogue analysis main task module, a lesson interaction analysis main task module, and a curriculum standard implementation analysis main task module as the main tasks of collaborative training, and conducting supervised fine-tuning training; the question-and-answer dialogue analysis main task module is used to respond to the question-and-answer dialogue analysis command; the lesson interaction analysis main task module is used to respond to the lesson interaction analysis command; and the curriculum standard implementation analysis main task module is used to respond to the curriculum standard implementation analysis command.

2. The steps include inputting the input text and the lesson objective information into a predetermined model, the predetermined model processing the input text and the lesson objective information in accordance with the objective analysis command, and outputting the evaluation results of the lesson to be evaluated, If it is confirmed that the received objective analysis command is a question-and-answer dialogue analysis command, the input target text and the teaching objective information are input into a predetermined model, the predetermined model responds to the question-and-answer dialogue analysis command, executes the question-and-answer dialogue analysis main task module based on the input target text and the teaching objective information, and outputs the evaluation results of the evaluation target lesson in the question-and-answer dialogue analysis main task module. If it is confirmed that the received objective analysis command is a lesson interaction analysis command, the input target text and the teaching objective information are input into a predetermined model, the predetermined model responds to the lesson interaction analysis command, executes the lesson interaction analysis main task module based on the input target text and the teaching objective information, and outputs the evaluation results of the lesson to be evaluated in the lesson interaction analysis main task module. The method according to claim 1, characterized in that, upon confirming that the received objective analysis command is a curriculum standard implementation analysis command, input the input target text and the curriculum standard information into a predetermined model, the predetermined model responds to the curriculum standard implementation analysis command, executes the curriculum standard implementation analysis main task module based on the input target text and the curriculum standard information, and outputs the evaluation results of the evaluation target class in the curriculum standard implementation analysis main task module.

3. If it is confirmed that the received objective analysis command is a question-and-answer dialogue analysis command, the input target text and the teaching objective information are input into a predetermined model, the predetermined model responds to the question-and-answer dialogue analysis command, executes the question-and-answer dialogue analysis main task module based on the input target text and the teaching objective information, and outputs the evaluation results of the evaluation target lesson in the question-and-answer dialogue analysis main task module, If it is confirmed that the received objective analysis command is a question-and-answer dialogue analysis command, the predetermined model is confirmed to respond to the question-and-answer dialogue analysis command, and the predetermined model is set to wait for the execution of objective evaluation tasks in each type of evaluation dimension in the question-and-answer dialogue analysis main task module. The steps to determine the objective evaluation task, The method according to claim 2, characterized in that it includes the steps of inputting the input target text and the teaching objective information into a predetermined model, and having the question-and-answer dialogue analysis main task module execute the objective evaluation task based on the input target text and the teaching objective information, and outputting the evaluation results of the evaluation target class.

4. The aforementioned main task module for question-and-answer dialogue analysis includes an overall evaluation subtask module and a local evaluation subtask module. The aforementioned step of determining the objective evaluation task is, A step of obtaining a first type of evaluation prompt command, the first type of evaluation prompt command is used to prompt a predetermined model to output an evaluation result of the lesson under evaluation according to an evaluation dimension of question effectiveness analysis, the evaluation dimension of question effectiveness analysis includes an evaluation of the effectiveness of the teacher's questions in the lesson under evaluation, A step in which, in response to the first type of evaluation prompt command, it is determined that a first overall subtask in the overall evaluation subtask module and a first local subtask in the local evaluation subtask module jointly constitute a target evaluation task, wherein the first overall subtask is used to overall evaluate the overall effectiveness of all of the teacher's questions in the lesson being evaluated, and the first local subtask is used to specifically determine the effectiveness of each of the teacher's questions in the lesson being evaluated. and / or, A step of obtaining a second type of evaluation prompt command, the second type of evaluation prompt command being used to prompt a predetermined model to output an evaluation result of the lesson being evaluated according to an evaluation dimension of knowledge objectives, the evaluation dimension of knowledge objectives including an evaluation of the relevance between the teacher's questions and the teaching objectives in the lesson being evaluated, A step in which, in response to the second type of evaluation prompt command, it is determined that the second overall subtask in the overall evaluation subtask module and the second local subtask in the local evaluation subtask module jointly constitute an objective evaluation task, wherein the second overall subtask is used to evaluate the overall pre-set knowledge objectives in the lesson being evaluated, and the second local subtask is used to specifically evaluate the degree of correlation between the teacher's questions and the teaching objectives in the lesson being evaluated. and / or, A step of obtaining a third type of evaluation prompt command as the target evaluation task, wherein the third type of evaluation prompt command is used to prompt a predetermined model to output an evaluation result of the lesson to be evaluated according to an evaluation dimension of question avoidance, the evaluation dimension of question avoidance includes a step of identifying teacher-avoidable questions in the lesson to be evaluated, A step in which, in response to the third type of evaluation prompt command, it is determined that a first local subtask and a second local subtask in the local evaluation subtask module jointly constitute an objective evaluation task, wherein the first local subtask is used to specifically determine the effectiveness of each question asked by the teacher in the lesson being evaluated, and the second local subtask is used to specifically evaluate the degree of correlation between the teacher's questions and the learning objectives in the lesson being evaluated. and / or, A step of obtaining a fourth type of evaluation prompt command as the target evaluation task, wherein the fourth type of evaluation prompt command is used to prompt a predetermined model to output an evaluation result of the lesson to be evaluated according to an evaluation dimension of question optimization, the evaluation dimension of question optimization includes finding teacher-optimizable questions in the lesson to be evaluated, The method according to claim 3, characterized in that, in response to a fourth type of evaluation prompt command, it is determined that a third global subtask in the global evaluation subtask module and a third local subtask in the local evaluation subtask module jointly constitute a target evaluation task, wherein the third global subtask is used to propose an overall optimization of the teacher's questions in the lesson being evaluated, and the third local subtask is used to optimize and rewrite the teacher's questions in the lesson being evaluated.

5. The method according to any one of claims 1 to 3, wherein the question-and-answer dialogue analysis main task module includes an overall evaluation subtask module and a local evaluation subtask module, and the predetermined model, in the process of performing supervised fine-tuning training on the question-and-answer dialogue analysis main task module, processes the lesson text containing the teaching objectives and teacher questions in the educational scene corpus according to the overall evaluation subtask module and the local evaluation subtask module, respectively, and outputs an evaluation result for the lesson text.

6. The method according to claim 5, wherein the overall evaluation subtask module includes a first overall subtask, a second overall subtask, and a third overall subtask, the first overall subtask being used in supervised fine-tuning training to overall evaluate the overall effectiveness of all of the teacher's questions in the course text, the second overall subtask being used in supervised fine-tuning training to overall evaluate the pre-set knowledge objectives in the course text, and the third overall subtask being used in supervised fine-tuning training to propose an overall optimization of the teacher's questions in the course text; the local evaluation subtask module includes a first local subtask, a second local subtask, and a third local subtask, the first local subtask being used in supervised fine-tuning training to specifically determine the effectiveness of each of the teacher's questions in the course text, the second local subtask being used in supervised fine-tuning training to specifically evaluate the correlation between the teacher's questions and the teaching objectives in the course text, and the third local subtask being used in supervised fine-tuning training to optimize and rewrite the teacher's questions in the course text.

7. The above step of obtaining lesson data and lesson objective information for the lessons to be evaluated is: Steps to obtain lesson data for the course to be evaluated, The method according to any one of claims 1 to 4, characterized by including the steps of querying a pre-configured database for lesson objective information that matches the lesson data, or inputting the lesson data into a pre-configured chapter matching model and causing the chapter matching model to output lesson objective information that matches the lesson data.

8. The step of converting the aforementioned lesson data into input text is: The steps include: converting the aforementioned lesson data into text to obtain the lesson text, The steps include: extracting the teacher's question text from the aforementioned lesson text, The method according to any one of claims 1 to 4, characterized by comprising the step of using the aforementioned lesson text and the aforementioned teacher's question text as input text.

9. The step of extracting the teacher's question text from the aforementioned lesson text is: The steps include identifying the dialogue character who appears most frequently in the aforementioned lesson text, The step of determining the dialogue character with the most dialogues as the teacher character, The steps include: extracting all dialogue content of the aforementioned teacher character, The method according to claim 8, further comprising the step of screening teacher question texts from all of the above dialogue content.

10. The step of screening the teacher's question text from all the aforementioned dialogue content is: The steps include identifying the basic question sentences from all of the above dialogue content, The method according to claim 9, characterized by comprising the step of performing a noise reduction operation on the basic question text to obtain the teacher's question text.

11. It is a teaching evaluation device, Used to acquire lesson data and lesson objective information for the lessons to be evaluated, the lesson data includes audio data and / or video data, and the lesson objective information includes a data acquisition module that includes teaching objective information and curriculum standard information. A data conversion module for converting the aforementioned lesson data into input text, Used to receive objective analysis commands for the aforementioned evaluation subject class, the objective analysis command includes a command receiving module that includes a question-and-answer dialogue analysis command, a class interaction analysis command, and a curriculum standard implementation analysis command, A lesson evaluation device characterized by comprising: inputting the input target text and the lesson objective information into a predetermined model, the predetermined model processing the input target text and the lesson objective information in accordance with the objective analysis command, and outputting the evaluation results of the lesson to be evaluated; the predetermined model is obtained by using an educational scene corpus as a training sample, using a question-and-answer dialogue analysis main task module, a lesson interaction analysis main task module, and a curriculum standard implementation analysis main task module as the main tasks of collaborative training, and performing supervised fine-tuning training; the question-and-answer dialogue analysis main task module is used to respond to the question-and-answer dialogue analysis command, the lesson interaction analysis main task module is used to respond to the lesson interaction analysis command, and the curriculum standard implementation analysis main task module is used to respond to the curriculum standard implementation analysis command.

12. An electronic device comprising a processor, memory, and a program or instruction stored in the memory and executable by the processor, wherein when the program or instruction is executed by the processor, the steps of the lesson evaluation method described in any one of claims 1 to 10 are realized.

13. A readable storage medium that stores a program or instruction, and when the program or instruction is executed by a processor, the steps of the lesson evaluation method described in any one of claims 1 to 10 are realized.

Images