Courseware script processing method and device, electronic equipment and computer storage medium

Abstract

The invention provides a courseware script processing method and device, and relates to the technical fields of artificial intelligence, natural language processing, computer vision and the like. The specific implementation scheme is as follows: generating a courseware script to be evaluated based on teaching demand information of an education subject; on the basis of the courseware script, role prompt words corresponding to different script evaluation roles are generated for the at least two evaluation large models; sending the courseware scripts and the role cues to the at least two evaluation large models, so that each evaluation large model stands at the site of the script evaluation role corresponding to the role cue to evaluate the courseware scripts, and obtaining courseware evaluation information output by the at least two evaluation large models; based on the courseware evaluation information output by the at least two evaluation large models, obtaining comprehensive evaluation feedback information; and based on the comprehensive evaluation feedback information, optimizing the courseware script to obtain an optimized courseware script.

Description

Technical Field

[0001] The present disclosure belongs to the field of educational information technology, particularly to the fields of artificial intelligence, natural language processing, computer vision, etc. In particular, it provides a courseware script processing method and device, an electronic device, and a computer-readable storage medium. Background Art

[0002] The traditional courseware script production process relies heavily on manual experience, expertise, and time. Teachers often need to expend considerable effort on content collection, structural design, language organization, and the conception of teaching activities. This process is not only time-consuming and labor-intensive, but the quality of the resulting scripts is often inconsistent, making it difficult to guarantee teaching effectiveness. While template-based script generation can improve efficiency to a certain extent, it often lacks flexibility, easily produces homogenized content, and lacks support for interactive design. While early computer-assisted instruction and multimedia courseware production introduced multimedia elements, they still had limitations in adaptability, depth of interaction, and production professionalism.

[0003] In recent years, artificial intelligence, particularly large language models, has begun to be applied to assist in the generation and evaluation of teaching content. For example, large language models have been used to generate preliminary syllabi, draft lesson plans, and courseware scripts. However, existing AI-assisted methods still suffer from numerous shortcomings, including low content generation quality, poor controllability, limited evaluation, and inefficient iterative optimization. Summary of the Invention

[0004] The present disclosure provides a courseware script processing method and device, an electronic device, and a computer-readable storage medium.

[0005] According to the first aspect, a courseware script processing method is provided, which includes: generating a courseware script to be evaluated based on the teaching demand information of the educational subject; generating role prompt words corresponding to different script evaluation roles for at least two evaluation large models based on the courseware script; sending the courseware script and the role prompt words to at least two evaluation large models, so that each evaluation large model evaluates the courseware script from the standpoint of the script evaluation role corresponding to its own role prompt word, and obtains courseware evaluation information output by at least two evaluation large models; obtains comprehensive evaluation feedback information based on the courseware evaluation information output by at least two evaluation large models; and optimizes the courseware script based on the comprehensive evaluation feedback information to obtain an optimized courseware script.

[0006] According to the second aspect, a courseware script processing device is provided, which includes: a script generation unit, configured to generate a courseware script to be evaluated based on the teaching demand information of the education subject; a prompt word generation unit, configured to generate role prompt words corresponding to different script evaluation roles for at least two evaluation large models based on the courseware script; a sending unit, configured to send the courseware script and role prompt words to at least two evaluation large models, so that each evaluation large model evaluates the courseware script from the standpoint of the script evaluation role corresponding to its own role prompt word, and obtains courseware evaluation information output by at least two evaluation large models; an evaluation unit, configured to obtain comprehensive evaluation feedback information based on the courseware evaluation information output by at least two evaluation large models; an optimization unit, configured to optimize the courseware script based on the comprehensive evaluation feedback information, and obtain an optimized courseware script.

[0007] According to a third aspect, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method described in any implementation manner of the first aspect.

[0008] According to a fourth aspect, a non-transitory computer-readable storage medium storing computer instructions is provided, where the computer instructions are used to cause a computer to execute the method as described in any implementation of the first aspect.

[0009] The courseware script processing method and device provided by the embodiment of the present disclosure firstly generates a courseware script to be evaluated based on the teaching demand information of the education subject; secondly, based on the courseware script, role prompt words corresponding to different script evaluation roles are generated for at least two evaluation large models; thirdly, the courseware script and role prompt words are sent to at least two evaluation large models, so that each evaluation large model evaluates the courseware script from the standpoint of the script evaluation role corresponding to its own role prompt word, and obtains the courseware evaluation information output by at least two evaluation large models; then, based on the courseware evaluation information output by at least two evaluation large models, comprehensive evaluation feedback information is obtained; finally, based on the comprehensive evaluation feedback information, the courseware script is optimized to obtain an optimized courseware script. Thus, at least two evaluation large models are used to set script evaluation roles for them respectively, which solves the problems of strong subjectivity and single evaluation dimension of traditional script evaluation methods, and realizes the automatic optimization of courseware scripts through precise engineering design and collaborative workflow, thereby improving the optimization efficiency of scripts.

[0010] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The accompanying drawings are provided to facilitate a better understanding of the present invention and do not constitute a limitation of the present disclosure.

[0012] Figure 1 is a flow chart of an embodiment of the method for processing courseware scripts according to the present disclosure;

[0013] Figure 2 It is an interface display diagram for evaluating multiple large evaluation models and comprehensive evaluation feedback information in this disclosure;

[0014] Figure 3 It is a schematic diagram of the interaction between various subjects in the script processing method of this open courseware;

[0015] Figure 4 It is a structural diagram of an embodiment of the open courseware script processing device;

[0016] Figure 5 It is a block diagram of an electronic device used to implement the courseware script processing method of the embodiment of the present disclosure. DETAILED DESCRIPTION

[0017] Unless expressly stated otherwise, throughout the specification and claims, the term "comprise" or variations such as "include" or "comprising", etc., will be understood to include the stated elements or components but not to exclude other elements or other components.

[0018] The technical solutions of the present disclosure are described below through specific examples. It should be understood that one or more steps mentioned in the present disclosure do not exclude the existence of other methods and steps before and after the combination step, or other methods and steps may be inserted between these explicitly mentioned steps. It should also be understood that these examples are only used to illustrate the present disclosure and are not used to limit the scope of the present disclosure. Unless otherwise specified, the numbering of each method step is only for the purpose of identifying each method step, and does not limit the order of arrangement of each method or limit the scope of implementation of the present disclosure. Changes or adjustments in their relative relationships can also be regarded as the scope of implementation of the present disclosure without substantial changes in the technical content.

[0019] The sources of the raw materials and instruments used in the examples are not particularly limited and can be purchased from the market or prepared according to conventional methods known to those skilled in the art.

[0020] The existing script evaluation method using a single AI (Artificial Intelligence) model has the following shortcomings:

[0021] The scripts generated by a single AI model may contain factual errors, illogical logic, deviation from teaching objectives, or lack of in-depth teaching understanding (such as insufficient understanding of the cognitive characteristics of students in a specific stage of education). The generated content may also be relatively rigid, with limited innovation and interest. Existing AI assessment tools are still insufficient in understanding the depth of teaching content, the comprehensiveness of assessment dimensions (such as it is difficult to balance scientificity, fun, and interactivity), and the provision of specific and actionable feedback. In particular, it is difficult to simultaneously simulate the control of professionalism by teaching experts and the true feelings of students about their acceptance of the content. Even if evaluation feedback is obtained, how to efficiently integrate the feedback into the modification and iteration of the courseware script to form an effective optimization closed loop is still a challenge currently faced.

[0022] Therefore, how to effectively utilize AI technology, especially the collaborative capabilities of multiple large-scale language models, to achieve high-quality automatic generation of courseware scripts, multi-dimensional intelligent evaluation, and efficient iterative optimization, is a technical problem that needs to be urgently solved in the current field of educational technology.

[0023] In view of the defects in traditional technologies, the present disclosure proposes a courseware script processing method. Figure 1 A process 100 according to an embodiment of the disclosed courseware script processing method is shown. The courseware script processing method includes the following steps:

[0024] Step 101: Generate a courseware script to be evaluated based on the teaching demand information of the education subject.

[0025] In this embodiment, the educational subject is the subject of courseware execution, including: educators (such as teachers, educational institutions, course designers, etc.) and educational terminals, among which educators send teaching demand information by operating the educational terminal. Educational demand information refers to a series of requirements and goals proposed by the educational subject when designing and developing courseware scripts. This information is the key factor to ensure that the courseware script can meet the teaching objectives, adapt to students' needs and effectively promote learning.

[0026] In this embodiment, the above step 101 includes: determining the structural information of the courseware script and basic courseware information such as teaching objectives, course outline, and learning background based on teaching demand information; performing data preprocessing (such as data cleaning and standardization) on the structural information of the script and the basic courseware information to obtain preprocessed data; designing a template for the courseware script based on the structural information in the preprocessed data; writing code to fill in the above template based on the preprocessed data to obtain the courseware script to be evaluated. Specifically, the above step 101 includes: generating a course content text based on the basic courseware information in the preprocessed data; selecting appropriate content and teaching methods based on the basic courseware information, and dynamically binding the course content text to the template according to the content and teaching methods.

[0027] In this embodiment, the courseware script to be evaluated can be implemented through a large model, such as Figure 2 As shown, the creative model LLM_1 is completed and generates the courseware script to be evaluated.

[0028] Step 102: Based on the courseware script, generate role prompt words corresponding to different script evaluation roles for at least two evaluation models.

[0029] In this embodiment, it is necessary to clarify the core content and teaching objectives of the courseware script, and analyze its applicable teaching scenarios and audience groups. Then, based on this information, a unique role perspective is designed for each evaluation model, such as the "student user" and "education expert" roles. For the "student user" role, the role prompt words can be: "You are a middle school student who is studying this course. Please evaluate the courseware script's ease of understanding, fun, and whether it can help you better master the knowledge points based on your comprehension ability and learning interest." For the "education expert" role, the role prompt words can be: "You are a senior education expert with rich teaching experience and course design knowledge. Please conduct a professional evaluation of the courseware script from the aspects of teaching logic, knowledge point coverage, teaching goal achievement, and innovation." By providing clear role prompt words for each evaluation model, they are guided to evaluate the courseware script from different perspectives, thereby obtaining more comprehensive and multi-dimensional evaluation results.

[0030] Optionally, step 102 further includes: determining a target academic stage or target subject based on the courseware script; determining role prompt word generation rules based on the target academic stage or target subject; and generating role prompt word generation rules for at least two large assessment models corresponding to different script assessment roles based on the role prompt word generation rules. Determining the role prompt word generation rules based on the target academic stage or target subject includes: determining a learning object corresponding to the target academic stage based on the target academic stage; determining a role prompt word generation rule for the learning object based on the learning object; and / or determining the subject characteristics of the target subject based on the target subject; and determining a role prompt word generation rule for the learning object based on the subject characteristics.

[0031] The above-mentioned rules for generating role prompts for learning objects can include the following scenarios: For example, if the target learning stage is elementary school, and the learning objects for elementary school are elementary school students, the role prompt generation rules for elementary school students include: the language must be easy to understand and consistent with the cognitive level of elementary school students; the courseware script must be interesting and able to attract elementary school students' attention; and the presentation of basic knowledge must be focused on whether it conforms to the elementary school students' cognitive process from concrete to abstract. For example, role prompts generated using this role prompt generation rule include: As an elementary school Chinese teacher, I will pay attention to whether the stories in the courseware script are vivid and interesting, attracting children like a wonderful fairy tale. It is also necessary to check whether the explanation of new characters and words is simple and easy to understand, just like telling children the names of new friends, so that they can easily remember them.

[0032] For middle school students, the role prompt generation rules include: Language can appropriately use some subject-specific terminology, but it must also ensure that students are comprehensible, as middle school students already have a certain foundation in subject knowledge and are able to accept a certain level of academic content; The courseware script should be logically and systematically organized. The middle school curriculum is relatively complex, so the courseware script needs to have a clear knowledge structure. For example, in a physics courseware script, knowledge points in mechanics should be presented sequentially from basic concepts to complex applications; and attention should be paid to whether the courseware script can guide students' thinking and exploration, cultivating their thinking skills. For example, the role prompt generated by this role prompt generation rule includes: "As a middle school history teacher, I carefully check whether the historical events in the courseware script are presented in chronological order and causally, just like constructing the framework of a historical story. I also check whether it can guide students to think about the reasons behind historical events, as if allowing them to solve historical mysteries on their own."

[0033] For university students, the role prompt generation rules include: using professional and accurate academic language. University students already possess strong academic skills and are able to understand and appreciate professional expressions. A key focus is on evaluating the depth and cutting-edge nature of courseware scripts. University courses often involve in-depth exploration of professional knowledge and cutting-edge disciplinary developments, and courseware scripts need to reflect these. For example, computer science courseware scripts should include the latest programming techniques and algorithmic theory. Furthermore, attention is paid to whether courseware scripts can promote students' independent learning and research abilities. For example, courseware scripts can provide guidance on research topics, allowing students to independently research and solve problems. For example, role prompts generated using this role prompt generation rule include: "As a university economics teacher, I will rigorously examine the accuracy of the theoretical exposition in a courseware script and whether it can deeply analyze complex economic models, just like conducting a high-level academic seminar. Furthermore, I will examine whether it introduces the latest economic cases, allowing students to engage with cutting-edge economic developments."

[0034] In this embodiment, different subjects have corresponding subject characteristics, and corresponding role prompt word generation rules can be determined based on the subject characteristics.

[0035] For example, the subject characteristics of Chinese language are: the appeal of text, the cultivation of students' reading and writing skills, and the requirements for grammatical vocabulary; the above-mentioned rules for generating role prompt words for learning objects based on subject characteristics include: for the appeal of text, the rules for generating role prompt words for learning objects include: evaluating the literary quality and the elegance of language expression of the courseware script; paying attention to the guidance of the courseware script on reading comprehension and writing skills; and detecting whether the courseware script conforms to language standards.

[0036] For example, the characteristics of mathematics include strict logical order, multi-angle problem analysis, and basic subject requirements. The rules for generating role prompts for learning objects based on these characteristics include: evaluating the logical rigor of the courseware script. Mathematical knowledge is interconnected, and the courseware script must follow a strict logical order; paying attention to whether the courseware script provides diverse problem-solving methods and ideas. Mathematics teaching encourages students to think about problems from different perspectives, and the courseware script should reflect this; and checking the accuracy of the mathematical symbols and formulas in the courseware script.

[0037] Step 103: Send the courseware script and role prompt words to at least two evaluation large models, so that each evaluation large model evaluates the courseware script from the perspective of the script evaluation role corresponding to its role prompt word, and obtains courseware evaluation information output by at least two evaluation large models.

[0038] In this embodiment, the evaluation model is a model that evaluates the courseware script according to the prompt word requirements. Any two of the at least two evaluation models can simultaneously obtain the same role prompt word, so that the two evaluation models both evaluate the courseware script from the perspective of the same evaluation role; any two of the at least two evaluation models can also obtain different role prompt words. Figure 2 As shown, at least two large evaluation models include: evaluation large model LLM_2-1, evaluation large model LLM_2-2, evaluation large model LLM_3-1, and evaluation large model LLM_3-2, and multi-role evaluation is achieved through multiple models.

[0039] exist Figure 2 In the example, the evaluation models LLM_2-1 / 2 and LLM_3-1 / 2 are described as independently performing dual-role evaluations, each assuming the roles of "expert" and "student." This can be understood as a single-model, dual-role approach: a prompt containing both the "expert role" and "student role" is sent simultaneously to the evaluation model LLM_2. This means that the evaluation model LLM_2-1 evaluates from the perspective of the expert, while the evaluation model LLM_2-2 evaluates from the perspective of the student. The evaluation model LLM_2-1 / 2 then outputs two evaluations simultaneously. The same process is repeated for the evaluation model LLM_3.

[0040] Optionally, there is a second implementation method for setting up the big model - one model, one role: send the "student role" prompt word to the evaluation big model LLM_A to obtain "student evaluation A". Send the "expert role" prompt word to the evaluation big model LLM_B to obtain the courseware evaluation information of "expert evaluation A". Send the "student role prompt word" to the evaluation big model LLM_C to obtain the courseware evaluation information of "student evaluation B". Send the "expert role prompt word" to the evaluation big model LLM_D to obtain "expert evaluation B". The evaluation big model LLM_A, evaluation big model LLM_B, evaluation big model LLM_C, and evaluation big model LLM_D here can be two models (for example, the evaluation big model LLM_A and the evaluation big model LLM_C are the same model, and the evaluation big model LLM_B and the evaluation big model LLM_D are another model), or they can be four different models to achieve maximum diversity.

[0041] In this embodiment, in order to conduct a comprehensive and multi-angle evaluation of the courseware script, the courseware script and role prompts are first sent to at least two different evaluation big models. The role prompts are used to clarify the role played by each evaluation big model in the evaluation process, such as "student", "teacher" or "education expert", etc. Each evaluation big model analyzes and evaluates the courseware script from the perspective of the role assigned to it, that is, from the standpoint of the role. For example, a big model that evaluates in the role of "student" will focus on the comprehensibility, fun and interactivity of the script; while a big model that evaluates in the role of "teacher" may pay more attention to the teaching logic, the completeness of the knowledge points and the achievement of teaching objectives. In this way, each evaluation big model will output evaluation information for the courseware script, which may include analysis of the advantages, disadvantages, improvement suggestions and potential problems of the script. Figure 2 As shown, the large-scale model evaluating the courseware using the "age-appropriate student" role provides evaluation information including: 1. Disincentive use of technical terminology; 2. Lack of practical examples; 3. Learning feels like a roller coaster ride; 4. Enigmatic task instructions. The large-scale model evaluating the courseware using the "education expert" role provides evaluation information including: 1. Meets the concept of holistic education; 2. Rich and diverse practical activities; 3. Highly innovative modular design. Clicking "Advantages" or "Confirm" on the display interface displays the strengths and weaknesses of the courseware evaluation information.

[0042] In this embodiment, the courseware evaluation information from different role perspectives is summarized to obtain the courseware evaluation information output by at least two large evaluation models. The courseware evaluation information from at least two large evaluation models provides a multi-dimensional reference basis for subsequent optimization and improvement.

[0043] Step 104: Obtain comprehensive evaluation feedback information based on the courseware evaluation information output by at least two large evaluation models.

[0044] In this embodiment, in order to obtain comprehensive evaluation feedback information, we first collect the evaluation information output by at least two large evaluation models for the courseware script. This evaluation information may include evaluations of the accuracy, logic, effectiveness of teaching methods, clarity of language expression, and other aspects of the courseware content, as well as their respective advantages, areas for improvement, actionable suggestions, and points of conflict. Next, we integrate and analyze the output results from these different large evaluation models, extract commonalities and differences, and comprehensively consider the evaluation focus and suggestions of each model to form a comprehensive comprehensive evaluation feedback information. This comprehensive evaluation feedback information will cover the overall performance of the courseware script, clearly point out its strengths and weaknesses, and provide specific improvement suggestions, providing systematic guidance for subsequent courseware optimization.

[0045] Optionally, the above step 104 includes: determining the evaluation role corresponding to each courseware evaluation information and the evaluation dimensions of each evaluation role, such as for the evaluation role of student, the evaluation dimensions include: content comprehensibility, degree of learning interest stimulation, and knowledge extensibility; for the evaluation role of expert, the evaluation dimensions include: curriculum standard compliance, teaching evaluation support, and teaching characteristics embodiment; using the evaluation dimensions of each evaluation information to evaluate the evaluation information to obtain the evaluation results of each evaluation information; according to the usage scenario and main audience of the courseware script, determine the importance of each evaluation dimension, based on the importance of each evaluation dimension, assign weights to the evaluation dimensions of each evaluation role, sort the courseware evaluation information output by at least two large evaluation models according to the weight of each evaluation dimension, and obtain comprehensive evaluation feedback information. A comprehensive and objective evaluation result can be given through the comprehensive evaluation feedback information. The above-mentioned importance of each evaluation dimension is determined based on the usage scenario and main audience of the courseware script. The following are: based on the grade of the courseware script, determine the evaluation role and evaluation dimensions of that grade; extract the courseware evaluation information of the evaluation role, and determine the importance of each evaluation dimension according to the importance of the evaluation role in that grade. For example, for elementary school Chinese courseware scripts: Teacher role: teaching goal alignment (30%), rationality of knowledge structure (20%), diversity of teaching methods (25%), resource utilization efficiency (25%); Student role: content comprehensibility (30%), degree of learning interest stimulation (40%), knowledge expansion (20%), and cultivation of higher-order thinking ability (10%); Education administrator role: curriculum standard compliance (40%), teaching effect evaluation support (30%), school teaching characteristics (20%), and teaching method diversity (10%). Based on the importance of each evaluation dimension, determine the importance of the corresponding courseware evaluation information, determine the main advantages, points for improvement, feasible modification suggestions of courseware evaluation information of different importance, and conflict points caused by the implementation of improvement points between each courseware evaluation information.

[0046] In this embodiment, the comprehensive evaluation feedback information can be the result obtained by analyzing the large model, such as Figure 2 The comprehensive large model LLM_4 shown here comprehensively analyzes the courseware evaluation information of at least two evaluation large models and provides comprehensive evaluation feedback information, such as Figure 4 The comprehensive evaluation feedback information includes: comprehensive modification suggestions from both parties, which include: 1. Making concept teaching more life-like 2. Optimizing learning rhythm and transition design 3. Refining practical activity guidelines.

[0047] Step 105: Optimize the courseware script based on the comprehensive evaluation feedback information to obtain an optimized courseware script.

[0048] In this embodiment, in order to optimize the courseware script based on the comprehensive evaluation feedback information, the comprehensive evaluation feedback information is first carefully analyzed, focusing on the main advantages, points for improvement, operational modification suggestions and conflict points. Based on these feedback contents, specific optimization strategies are formulated, such as retaining and strengthening the advantages, making targeted corrections to the shortcomings, adjusting the content structure or language expression according to the suggestions, and resolving conflict points to ensure the logical coherence and consistency of the script. Then, on the basis of the courseware script, these optimization measures are implemented one by one, and the script content is modified and improved. After the optimization is completed, the script is reviewed to ensure that all feedback suggestions have been properly handled and the script quality has been improved, and finally the optimized courseware script is obtained to make it more in line with the teaching objectives and actual needs.

[0049] The courseware script processing method provided by the embodiment of the present disclosure first generates a courseware script to be evaluated based on the teaching demand information of the educational subject; secondly, based on the courseware script, role prompt words corresponding to different script evaluation roles are generated for at least two evaluation large models; thirdly, the courseware script and role prompt words are sent to at least two evaluation large models, so that each evaluation large model evaluates the courseware script from the standpoint of the script evaluation role corresponding to its own role prompt word, and obtains courseware evaluation information output by at least two evaluation large models; then, based on the courseware evaluation information output by at least two evaluation large models, comprehensive evaluation feedback information is obtained; finally, based on the comprehensive evaluation feedback information, the courseware script is optimized to obtain an optimized courseware script.

[0050] In some optional implementations of the present disclosure, the above method further includes: sending the optimized courseware script to the education subject; responding to detecting the revised script feedback from the education subject; and obtaining the target courseware script based on the revised script and historical evaluation feedback information.

[0051] In this optional implementation, the target courseware script is the script that best meets the teaching requirement information, and the target courseware script is the courseware script obtained after being modified by the education subject.

[0052] In this optional implementation, the generated target courseware script ensures that the teacher's modifications are "fine-tuned" based on the optimized courseware script, rather than inadvertently overturning previous effective improvements. A second purpose is to make the re-evaluation more targeted, focusing on the teacher's modifications and determining whether these modifications resolve historical issues or introduce new ones.

[0053] When the execution subject on which the courseware script processing method is running receives the "revised script" (such as v3.0) submitted by the teacher, it will not simply and crudely use it to overwrite the previously optimized courseware script (v2.0). Specifically, based on the revised script and historical evaluation feedback information, the target courseware script obtained includes:

[0054] First, a text comparison is performed on the optimized courseware script and the revised script to determine the modification position of the revised script relative to the optimized courseware script. Then, based on the modification position of the courseware script, a correlation analysis is performed on the historical evaluation feedback information to generate a more intelligent evaluation instruction with rich context and send it to at least two evaluation large models (such as the above-mentioned evaluation large model LLM_2 and evaluation large model LLM_3). At least two evaluation large models output a new round of more targeted analysis feedback reports based on this "context-aware" instruction. Finally, the creation model (the large model of the original creation courseware script, such as LLM_1) receives the teacher's revised script and analysis feedback report, performs final fusion and optimization, and generates the target courseware script (which can be called v4.0). This target courseware script not only retains the essence of the teacher's manual editing, but also ensures the closed loop and improvement of the overall quality through the re-examination of AI.

[0055] Optionally, the above-mentioned target courseware script obtained based on the revised script and historical evaluation feedback information may also include: after the optimized courseware script is sent to the education subject (such as a teacher or an education institution), the system will monitor the revised script fed back by the education subject in real time. After receiving the revised script fed back by the education subject, the system first performs a structured analysis of the revised content in the revised script through natural language processing technology, identifies key change points such as added, deleted, and modified chapters, knowledge point expression adjustments, and interactive link modifications; at the same time, retrieves historical evaluation feedback information from the historical evaluation feedback database (including student classroom interaction data, post-class test scores, eye tracking attention analysis and other multimodal data), and uses the system based on The cross-modal fusion algorithm of the attention mechanism conducts correlation analysis between the revised script and dimensions such as weak points in knowledge mastery, periods of distraction, and cold spots in interaction in historical assessment feedback information; multiple rounds of iterative optimization are carried out through the preset courseware quality evaluation model (including educational psychology indicators such as cognitive load theory and multimedia learning principles), and the target courseware script is finally generated. The target courseware script will retain 70% of the subjective teaching intentions in the revised script, and automatically embed 30% of optimization modules driven by historical data (such as dynamic visual splitting of difficult knowledge and enhanced interactive design of frequently wrong questions), and the educational basis and data support source of each adjustment are detailed in the script notes.

[0056] The method for obtaining the target courseware script provided by this optional implementation method obtains the target courseware script based on the revised script and historical evaluation feedback information. This allows the target courseware script to integrate the modifications of the educational subject and historical evaluation feedback information, thereby improving the reliability and accuracy of the target courseware script.

[0057] In a specific example, Figure 3 As shown, the authoring model LLM_1 is responsible for generating the courseware script to be evaluated (Courseware Script_v1.0) based on the initial input and the educational subject's prompts. It then iteratively optimizes the script after receiving the "Comprehensive Evaluation Information Feedback Report." The large evaluation model LLM_2, using the specific prompts of reviewer A, plays the roles of "education expert / subject expert" and "student of the corresponding grade," respectively, conducting a first round of dual-role evaluation of the courseware script generated by LLM_1 and outputting expert evaluation information A and student evaluation information A. The large evaluation model LLM_3 (or a model with a different type or slightly different prompts from LLM_2) also plays a dual-role role in the second round of independent evaluation of the script, outputting expert evaluation information B and student evaluation information B, aiming to provide diverse feedback and enhance evaluation robustness. Evaluation information integration module: It adopts a hybrid algorithm system based on the combination of rules and large models, responsible for integrating the four original evaluation reports of the evaluation large model LLM_2 and the evaluation large model LLM_3, performing semantic deduplication, dimension classification, consensus analysis, weight allocation (the student perspective is preset to have a higher weight than the expert perspective), conflicting evaluation processing, and finally using its built-in LLM to extract and generate structured comprehensive evaluation information feedback information. Figure 3 The courseware script processing method shown includes the following stages:

[0058] Phase 1: Generate the courseware script to be evaluated. Teachers input teaching requirements through an interface, which can include the courseware materials corresponding to the courseware script. The authoring model LLM_1 combines these requirements with the educational subject prompts, accesses the internal knowledge base, and generates the initial "Courseware Script_v1.0" containing structured content such as teaching objectives, key points and difficulties, steps, interactions, case studies, exercises, and a summary.

[0059] Phase 2: Multi-role, multi-model parallel evaluation: Courseware Script_v1.0 is distributed simultaneously to the evaluation master models LLM_2 and LLM_3. Each of these models independently completes the dual-role (expert and student) evaluation of Courseware Script_v1.0 based on its own specific prompt instruction set (including precise role-playing instructions, multi-dimensional evaluation instructions for experts and students, grade-level information adaptation, scoring system, output format requirements, etc.), outputting expert evaluation information A / B and student evaluation information A / B, respectively. The prompts in the evaluation master model LLM_3 can introduce subtle differences while maintaining consistency in core standards to generate more diverse feedback.

[0060] Phase 3: Integration and Refinement of Assessment Information: The assessment information integration module collects four sets of raw assessment information. Through its hybrid algorithm system, the LLM performs semantic deduplication and clustering, automatically annotates assessment information dimensions, assigns high priority to consensus assessment information, marks and analyzes conflicting assessment information (submitted to LLM_1 for subsequent judgment), and comprehensively considers it based on preset weights (student perspectives are weighted higher than expert perspectives). Ultimately, the processed information is refined into a plain text "Comprehensive Assessment Information Feedback" that includes key strengths, areas for improvement (by priority), specific and actionable modification suggestions, and conflicting points.

[0061] Phase 4: Iterative Optimization of the Courseware Script: Authoring Model LLM_1 receives Courseware Script_v1.0 and comprehensive evaluation feedback. LLM_1 leverages its natural language understanding capabilities to deeply analyze the comprehensive evaluation feedback and process each actionable feedback item, performing optimization operations such as adding or deleting content, adjusting the order, revising wording, replacing examples, and improving interactive design. When handling conflicting evaluation information, Authoring Model LLM_1 makes independent judgments based on pre-set optimization goals or weights. Ultimately, it outputs the "Optimized Courseware Script_v2.0."

[0062] Phase 5: Final Output and Optional Iteration: The optimized courseware script v2.0 is presented to the instructor in plain text via the user interface. The instructor can manually review the script and directly annotate or modify it to initiate a new iteration. After interpreting the instructor's input, the system combines the current script version with historical assessment records (for LLM evaluation reference) and restarts the process from Phases 2 to 4 to generate the target courseware script (v3.0, etc.) until the instructor is satisfied.

[0063] In some optional implementations of the present disclosure, the at least two evaluation large models include: a first large model and a second large model, and the role prompt words include: a first prompt word for the learning subject role, a second prompt word for the learning subject role, a first prompt word for the teaching evaluation role, and a second prompt word for the teaching evaluation role; sending the courseware script and the role prompt words to the at least two evaluation large models, so that each evaluation large model evaluates the courseware script from the standpoint of the script evaluation role corresponding to its role prompt word, and obtaining the courseware evaluation information output by at least two evaluation large models includes: sending the courseware script, the first prompt word for the learning subject role, and the first prompt word for the teaching evaluation role to the first large model, so that the first large model evaluates the courseware script from the standpoint of the learning subject role and the teaching evaluation role respectively. The script is evaluated to obtain the first learning subject role evaluation information and the first teaching evaluation role evaluation information output by the first large model; the courseware script, the second prompt word of the learning subject role and the second prompt word of the teaching evaluation role are sent to the second large model, so that the second large model evaluates the courseware script from the standpoints of the learning subject role and the teaching evaluation role respectively, and obtains the second learning subject role evaluation information and the second teaching evaluation role evaluation information output by the second large model; the above-mentioned courseware evaluation information based on the output of at least two evaluation large models obtains comprehensive evaluation feedback information, including: comprehensive evaluation feedback information obtained based on the first learning subject role evaluation information, the first teaching evaluation role evaluation information, the second learning subject role evaluation information and the second teaching evaluation role evaluation information.

[0064] In this optional implementation, when using at least two large evaluation models (the first large model and the second large model) to evaluate the courseware script, the system first combines and distributes the courseware script with the role prompt words: sends the courseware script, the first prompt words of the learning subject role, and the first prompt words of the teaching evaluation role to the first large model, so that it simulates the perspectives of the learning subject and the teaching evaluator respectively, and outputs the first learning subject role evaluation information (such as the difficulty of knowledge understanding, interactive experience) and the first teaching evaluation role evaluation information (such as teaching goal coverage, cognitive load rationality); at the same time, sends the courseware script, the second prompt words of the learning subject role (such as differentiated learning style description) and the second prompt words of the teaching evaluation role (such as curriculum standard compliance requirements) to the second large model to generate the second learning subject role evaluation information (such as contextual learning adaptability) and the second teaching evaluation role evaluation information (such as teaching method effectiveness).

[0065] In this optional implementation, semantic alignment technology can be used to merge similar role evaluation information. The specific steps are as follows: Sentiment polarity analysis and demand feature extraction are performed on learning subject information to form comprehensive learning subject information labeled with dimensions such as "cognitive curve smoothness" and "engagement incentives." For teaching evaluation information, a teaching logic tree is constructed through ontology reasoning to generate comprehensive teaching evaluation information with weights such as "knowledge system integrity" and "assessment strategy feasibility." Finally, a multi-objective optimization algorithm is used to perform weighted fusion on the clustered information based on preset dimension priorities (e.g., knowledge accessibility for beginners, depth of thinking for advanced courses), and output comprehensive evaluation feedback information with confidence scores and conflict markers.

[0066] The courseware script processing method provided in this embodiment, when at least two evaluation models include a first large model and a second large model, and the role prompt words include: a first prompt word for the learning subject role, a second prompt word for the learning subject role, a first prompt word for the teaching evaluation role, and a second prompt word for the teaching evaluation role, respectively sends the courseware script, the first prompt word for the learning subject role, and the first prompt word for the teaching evaluation role to the first large model, and sends the courseware script, the second prompt word for the learning subject role, and the second prompt word for the teaching evaluation role to the second large model, integrates the role evaluation information output by the first large model, and obtains comprehensive evaluation feedback information. By integrating multiple information, it is ensured that the optimization suggestions meet both the learner experience and the teaching professionalism requirements, thereby improving the reliability of the comprehensive evaluation feedback information.

[0067] In some optional implementations of the present disclosure, the above-mentioned comprehensive evaluation feedback information based on the first learning subject role evaluation information, the first teaching evaluation role evaluation information, the second learning subject role evaluation information and the second teaching evaluation role evaluation information includes: merging the first learning subject role evaluation information and the second learning subject role evaluation information to obtain learning subject comprehensive information; merging the first teaching evaluation role evaluation information and the second teaching evaluation role evaluation information to obtain teaching evaluation comprehensive information; deduplicating and clustering the learning subject comprehensive information and the teaching evaluation comprehensive information to obtain learning subject processing information and teaching evaluation processing information; labeling the learning subject processing information and the teaching evaluation processing information with different evaluation dimensions and weights to obtain comprehensive evaluation feedback information.

[0068] In this optional implementation, when integrating feedback from multiple large evaluation models, the system first uses semantic alignment technology to merge similar evaluation data. The learning subject role evaluation information generated by the first and second large models (e.g., "first learning subject role evaluation information" and "second learning subject role evaluation information") is normalized to extract key evaluation dimensions (e.g., difficulty of knowledge comprehension, learning interest, etc.). This information is then eliminated through an attention-based text fusion model to generate structured learning subject comprehensive information. Simultaneously, the teaching evaluation role evaluation information from the two models (e.g., teaching objective alignment, classroom interaction feasibility, etc.) is logically correlated and merged into comprehensive teaching evaluation information. Next, the system uses a hierarchical clustering algorithm (e.g., DBSCAN) to deduplicate the comprehensive information, merging semantically similar feedback items (e.g., "Insufficient number of examples" and "Practice session needs improvement" into the same category). Clustering topics are then annotated based on the educational knowledge graph to form de-redundant learning subject processing information and teaching evaluation processing information.

[0069] In this optional implementation method, the above-mentioned labeling of different evaluation dimensions and weights of the learning subject processing information and the teaching evaluation processing information to obtain comprehensive evaluation feedback information includes: based on the preset evaluation dimension framework (such as Bloom's educational objective taxonomy) and dynamic weight rules (such as the weight of "understanding difficulty" in historical data accounts for 30%), the processed information is multi-dimensionally labeled and weighted fused, and for non-conflicting learning subject processing information and teaching evaluation processing information of various weights under each evaluation dimension, the disposal method of the learning subject processing information and the teaching evaluation processing information is directly given; for conflicting learning subject processing information and teaching evaluation processing information of various weights under each evaluation dimension, the conflict status between the learning subject processing information and the teaching evaluation processing information is analyzed, and the corresponding disposal method is given to generate comprehensive evaluation feedback information including priority sorting, optimization suggestions and confidence scores.

[0070] This optional implementation provides a method for obtaining comprehensive evaluation feedback information, which divides the comprehensive information of the learning subject and the comprehensive information of the teaching evaluation, deduplicates and clusters the two, obtains the learning subject processing information and the teaching evaluation processing information, labels the learning subject processing information and the teaching evaluation processing information with different evaluation dimensions and weights, and provides a disposal method to ensure that the output results are both comprehensive and operational.

[0071] In some optional implementations of the present disclosure, the above-mentioned labeling of the learning subject processing information and the teaching evaluation processing information with different evaluation dimensions and weights to obtain comprehensive evaluation feedback information includes: labeling the learning subject processing information and the teaching evaluation processing information with consensus evaluation dimensions and conflicting evaluation dimensions to obtain learning subject labeling information and teaching evaluation labeling information; assigning a first weight to the learning subject labeling information, and assigning a second weight to the teaching evaluation labeling information, the first weight being greater than the second weight; adding the product of the learning subject labeling information and the first weight to the product of the teaching evaluation labeling information and the second weight to obtain comprehensive labeling data; evaluating the comprehensive labeling data to obtain comprehensive evaluation feedback information.

[0072] In this optional implementation, the learning subject's understanding of the courseware is more important than the teaching evaluation subject's understanding. To this end, a first weight is assigned to the learning subject's annotation information, and a second weight is assigned to the teaching evaluation annotation information, and the first weight is made greater than the second weight.

[0073] In this optional implementation, when generating comprehensive evaluation feedback information, the system first categorizes the learning subject processing information and the teaching evaluation processing information into dimensions. Through semantic analysis and rule matching, the two types of information are divided into consensus evaluation dimensions (such as the "clarity of knowledge point explanation" pointed out by both parties) and conflicting evaluation dimensions (such as the learning subject's feedback of "lack of interaction" while the teaching evaluation believes that the "rhythm is tight"), and are respectively annotated as structured learning subject annotation information and teaching evaluation annotation information. Then, based on the preset weight strategy (giving priority to learner experience), the system assigns a higher first weight (such as 0.7) to the learning subject annotation information and a lower second weight (such as 0.3) to the teaching evaluation annotation information, and generates comprehensive annotation data through weighted calculation (learning subject annotation information × 0.7 + teaching evaluation annotation information × 0.3).

[0074] In this optional implementation, the above-mentioned evaluation of the comprehensive annotated data obtains comprehensive evaluation feedback information including: using multi-level evaluation logic to directly adopt optimization suggestions for consensus dimensions, and arbitrating conflicting dimensions based on weight distribution results and historical data verification (such as the impact coefficient of "lack of interaction" on grades in past classes), and outputting comprehensive evaluation feedback information including priority ranking, correction plan and confidence score.

[0075] Optionally, the above-mentioned evaluation of the comprehensive annotation data to obtain comprehensive evaluation feedback information includes: inputting the comprehensive annotation data into the evaluation big model to obtain comprehensive evaluation feedback information output by the evaluation big model, wherein the comprehensive evaluation feedback information includes: main advantages, points for improvement, operational modification suggestions and conflict points.

[0076] In this optional implementation, to evaluate the comprehensive annotated data and obtain comprehensive evaluation feedback, the following specific implementation method can be used: First, the comprehensive annotated data is input into a pre-trained evaluation model. This evaluation model analyzes and evaluates the input data using a deep learning algorithm and outputs comprehensive evaluation feedback based on its internal evaluation mechanism and standards. Comprehensive evaluation feedback typically includes four aspects: first, key strengths, i.e., the parts of the annotated data that perform well, such as annotation accuracy and consistency; second, areas for improvement, which identify problems or deficiencies in the annotation, such as omissions or errors; third, actionable modification suggestions, which provide specific improvement suggestions and optimization directions for areas for improvement to help annotators make corrections; and fourth, conflict points, which identify contradictions or inconsistencies in the annotated data, such as inconsistent annotation results from different annotators on the same data. In this way, comprehensive evaluation feedback can provide a comprehensive and systematic reference for the quality assessment and optimization of annotated data.

[0077] The method for obtaining comprehensive evaluation feedback information provided by this optional implementation method assigns a first weight to the learning subject's annotation information and a second weight to the teaching evaluation annotation information, wherein the first weight is greater than the second weight. Comprehensive annotation data is obtained based on the first weight and the second weight. Comprehensive evaluation feedback information is obtained by evaluating the comprehensive annotation data, thereby ensuring that the final decision is both learner-centered and takes into account teaching professionalism, thereby improving the reliability of the comprehensive evaluation feedback information.

[0078] In some optional implementations of the present disclosure, the above-mentioned generation of the courseware script to be evaluated based on the teaching demand information of the educational subject includes: receiving the teaching demand information of the educational subject; determining the teaching knowledge base based on the teaching demand information; sending the teaching demand information, the teaching knowledge base and the prompt words of the educational subject to the execution big model, and obtaining the courseware script to be evaluated output by the execution big model.

[0079] In this optional implementation, the execution big model is a big model for generating courseware scripts. The education subject includes: teachers and terminals used by teachers. The prompt words of the education subject are prompts to generate courseware scripts from the perspective of the education subject.

[0080] In this optional implementation, in order to generate the courseware script to be evaluated, the teaching demand information provided by the education subject (such as a teacher or an educational institution) is first received. This information may include key elements such as teaching objectives, course themes, characteristics of the student group, and teaching duration. Then, based on this teaching demand information, a matching teaching knowledge base is screened and determined from existing educational resources. The knowledge base contains relevant teaching content, knowledge points, cases and other materials. Then, the teaching demand information, the determined teaching knowledge base, and the prompt words provided by the education subject (such as special instructions on the courseware style, key content, etc.) are integrated and sent to the execution model. The execution model uses its powerful generation capabilities, combined with the input information and prompt words, to output a courseware script to be evaluated that meets the needs of the education subject.

[0081] The method for generating courseware scripts provided by this optional implementation method, after receiving teaching demand information, can ensure the reliability and accuracy of the courseware script by sending the teaching demand information, teaching knowledge base and prompt words of the educational subject to the execution big model, and provide basic content for subsequent evaluation and optimization.

[0082] In some optional implementations of the present disclosure, the above-mentioned optimization of the courseware script based on the comprehensive evaluation feedback information to obtain the optimized courseware script includes: inputting the comprehensive evaluation feedback information into the execution big model to obtain the parsing feedback report output by the execution big model, including operational feedback items; decomposing the parsing feedback report into at least one optimization task; executing the operation corresponding to at least one optimization task on the basis of the courseware script; in response to the completion of at least one optimization task, checking the executed script to obtain the optimized courseware script.

[0083] In this optional implementation, the execution big model is the big model that generates the courseware script. After inputting the comprehensive evaluation feedback information into the execution big model, the execution big model can conduct an in-depth analysis of the comprehensive evaluation feedback information to determine the optimization tasks that can be performed on the generated courseware script.

[0084] In this optional implementation, in order to optimize the courseware script based on the comprehensive evaluation feedback information, the comprehensive evaluation feedback information is first input into the execution model, which will generate a parsing feedback report containing actionable feedback items based on the main advantages, points for improvement, actionable modification suggestions, and conflict points in the feedback information. Next, this parsing feedback report is decomposed into at least one specific optimization task, such as adjusting the content structure, correcting error information, optimizing language expression, etc. Then, based on the original courseware script, the operations corresponding to these optimization tasks are executed one by one to modify and improve the script in a targeted manner. When all optimization tasks in at least one optimization task are completed, the executed script is comprehensively checked to ensure that all optimization points have been implemented and the overall quality of the script has been improved, and finally an optimized courseware script is obtained to provide higher quality teaching materials for subsequent teaching activities.

[0085] Further references Figure 4 As an implementation of the methods shown in the above figures, the present disclosure provides an embodiment of a courseware script processing device. Figure 1 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0086] like Figure 4 As shown, the courseware script processing device 400 provided in this embodiment includes: a script generation unit 401, a prompt word generation unit 402, a sending unit 403, an evaluation unit 404, and an optimization unit 405. The script generation unit 401 can be configured to generate a courseware script to be evaluated based on the teaching needs information of the education subject. The prompt word generation unit 402 can be configured to generate role prompt words corresponding to different script evaluation roles for at least two evaluation models based on the courseware script. The sending unit 403 can be configured to send the courseware script and role prompt words to at least two evaluation models, so that each evaluation model evaluates the courseware script from the perspective of the script evaluation role corresponding to its role prompt word, thereby obtaining courseware evaluation information output by the at least two evaluation models. The evaluation unit 404 can be configured to obtain comprehensive evaluation feedback information based on the courseware evaluation information output by the at least two evaluation models. The optimization unit 405 may be configured to optimize the courseware script based on the comprehensive evaluation feedback information to obtain an optimized courseware script.

[0087] In this embodiment, the specific processing and technical effects of the script generation unit 401, prompt word generation unit 402, sending unit 403, evaluation unit 404, and optimization unit 405 in the courseware script processing device 400 can be referred to respectively. Figure 1The relevant descriptions of step 101, step 102, step 103, step 104 and step 105 in the corresponding embodiment are not repeated here.

[0088] In some embodiments of the present disclosure, the above-mentioned device also includes a revision unit (not shown in the figure), and the above-mentioned revision unit is configured to: send the optimized courseware script to the educational subject; respond to the detection of the revised script feedback from the educational subject; and obtain the target courseware script based on the revised script and historical evaluation feedback information.

[0089] In some embodiments of the present disclosure, the at least two evaluation large models include: a first large model and a second large model, and the role prompt words include: a first prompt word for the learning subject role, a second prompt word for the learning subject role, a first prompt word for the teaching evaluation role, and a second prompt word for the teaching evaluation role; the sending unit 403 is configured to: send the courseware script, the first prompt word for the learning subject role, and the first prompt word for the teaching evaluation role to the first large model, so that the first large model evaluates the courseware script from the perspectives of the learning subject role and the teaching evaluation role, and obtains the first learning subject role evaluation output by the first large model. The evaluation unit 404 is configured to obtain comprehensive evaluation feedback information based on the first learning subject role evaluation information, the first teaching evaluation role evaluation information, the second learning subject role evaluation information and the second teaching evaluation role evaluation information; the courseware script, the second prompt word of the learning subject role and the second prompt word of the teaching evaluation role are sent to the second largest model, so that the second largest model evaluates the courseware script from the standpoints of the learning subject role and the teaching evaluation role, and obtains the second learning subject role evaluation information and the second teaching evaluation role evaluation information output by the second largest model; the above-mentioned evaluation unit 404 is configured to obtain comprehensive evaluation feedback information based on the first learning subject role evaluation information, the first teaching evaluation role evaluation information, the second learning subject role evaluation information and the second teaching evaluation role evaluation information.

[0090] In some embodiments of the present disclosure, the above-mentioned evaluation unit 404 is further configured to: merge the first learning subject role evaluation information and the second learning subject role evaluation information to obtain the learning subject comprehensive information; merge the first teaching evaluation role evaluation information and the second teaching evaluation role evaluation information to obtain the teaching evaluation comprehensive information; deduplicate and cluster the learning subject comprehensive information and the teaching evaluation comprehensive information to obtain the learning subject processing information and the teaching evaluation processing information; label the learning subject processing information and the teaching evaluation processing information with different evaluation dimensions and weights to obtain comprehensive evaluation feedback information.

[0091] In some embodiments of the present disclosure, the evaluation unit 404 is further configured to: label the learning subject processing information and the teaching evaluation processing information according to consensus evaluation dimensions and conflict evaluation dimensions, to obtain learning subject labeled information and teaching evaluation labeled information; assign a first weight to the learning subject labeled information, and assign a second weight to the teaching evaluation labeled information, the first weight being greater than the second weight; multiply the learning subject labeled information by the first weight and add the product of the teaching evaluation labeled information and the second weight to obtain comprehensive labeled data; and evaluate the comprehensive labeled data to obtain comprehensive evaluation feedback information.

[0092] In some embodiments of the present disclosure, the script generation unit 401 is configured to: receive teaching demand information of an education subject; determine a teaching knowledge base based on the teaching demand information; and send the teaching demand information, the teaching knowledge base, and prompt words of the education subject to an execution large model to obtain a courseware script to be evaluated output by the execution large model.

[0093] In some embodiments of the present disclosure, the optimization unit 405 is configured to: input the comprehensive evaluation feedback information into the execution large model to obtain an analysis feedback report including operable feedback items output by the execution large model; decompose the analysis feedback report into at least one optimization task; perform an operation corresponding to the at least one optimization task on the basis of the courseware script; and in response to completion of execution of the at least one optimization task, check the executed script to obtain an optimized courseware script.

[0094] The courseware script processing apparatus provided by the embodiments of the present disclosure first generates a courseware script to be evaluated by the script generation unit 401 based on teaching demand information of an education subject; secondly, the sending unit 402 sends the courseware script, first prompt words of a learning subject role, and first prompt words of a teaching evaluation role to a first large model to obtain first learning subject role evaluation information and first teaching evaluation role evaluation information output by the first large model; thirdly, the sending unit 403 sends the courseware script, second prompt words of the learning subject role, and second prompt words of the teaching evaluation role to a second large model to obtain second learning subject role evaluation information and second teaching evaluation role evaluation information output by the second large model; then, the evaluation unit 404 obtains comprehensive evaluation feedback information based on the first learning subject role evaluation information, the first teaching evaluation role evaluation information, the second learning subject role evaluation information, and the second teaching evaluation role evaluation information; and finally, the optimization unit 405 optimizes the courseware script based on the comprehensive evaluation feedback information to obtain an optimized courseware script.

[0095] According to the embodiments of the present disclosure, the present disclosure further provides an electronic device, a readable storage medium, and a computer program product.

[0096] Figure 5A schematic block diagram of an example electronic device 500 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their modes are provided for example only and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0097] like Figure 5 As shown, the electronic device 500 includes a computing unit 501, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 502 or a computer program loaded from a storage unit 508 into a random access memory (RAM) 503. Various programs and data required for the operation of the electronic device 500 can also be stored in the RAM 503. The computing unit 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0098] Multiple components in the electronic device 500 are connected to the I / O interface 505, including: an input unit 506, such as a keyboard, a mouse, etc.; an output unit 507, such as various types of displays, speakers, etc.; a storage unit 508, such as a magnetic disk, an optical disk, etc.; and a communication unit 509, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 509 allows the electronic device 500 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0099] The computing unit 501 can be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 501 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 501 performs the various methods and processes described above, such as the courseware script processing method. For example, in some embodiments, the courseware script processing method can be implemented as a computer software program that is tangibly contained in a machine-readable medium, such as a storage unit 508. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 500 via the ROM 502 and / or the communication unit 509. When the computer program is loaded into the RAM 503 and executed by the computing unit 501, one or more steps of the courseware script processing method described above can be performed. Alternatively, in other embodiments, the computing unit 501 can be configured to execute the courseware script processing method in any other appropriate manner (e.g., by means of firmware).

[0100] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0101] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable courseware script processing device so that the program code, when executed by the processor or controller, causes the modes / operations specified in the flow chart and / or block diagram to be implemented. The program code can be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0102] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0103] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0104] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0105] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This is not limited herein.

[0106] The foregoing descriptions of specific exemplary embodiments of the present disclosure are for purposes of illustration and description. These descriptions are not intended to limit the present disclosure to the precise forms disclosed, and it is apparent that many variations and modifications are possible in light of the foregoing teachings. The exemplary embodiments have been selected and described for the purpose of explaining the specific principles of the present disclosure and their practical application, thereby enabling those skilled in the art to realize and utilize a variety of exemplary embodiments of the present disclosure and various options and modifications. The scope of the present disclosure is intended to be defined by the claims and their equivalents.

Claims

1. A courseware script processing method, characterized in that: The method comprises: Generate the courseware script to be evaluated based on the teaching demand information of the education subject; Based on the courseware script, generating role prompt words corresponding to different script evaluation roles for at least two evaluation models; Sending the courseware script and the role prompt words to the at least two evaluation large models, so that each evaluation large model evaluates the courseware script from the perspective of the script evaluation role corresponding to the role prompt word, and obtaining courseware evaluation information output by the at least two evaluation large models; Obtaining comprehensive evaluation feedback information based on the courseware evaluation information output by the at least two evaluation models; Based on the comprehensive evaluation feedback information, the courseware script is optimized to obtain an optimized courseware script.

2. The method according to claim 1, characterized in that The method further comprises: Sending the optimized courseware script to the education subject; a revised script in response to detecting feedback from the educational subject; Based on the revised script and historical evaluation feedback information, a target courseware script is obtained.

3. The method according to claim 1, characterized in that The at least two evaluation large models include: a first large model and a second large model; the role prompt words include: a first prompt word for a learning subject role, a second prompt word for a learning subject role, a first prompt word for a teaching evaluation role, and a second prompt word for a teaching evaluation role; the courseware script and the role prompt words are sent to the at least two evaluation large models, so that each evaluation large model evaluates the courseware script from the perspective of the script evaluation role corresponding to its role prompt word, and the courseware evaluation information output by the at least two evaluation large models includes: Sending the courseware script, the first prompt word of the learning subject role, and the first prompt word of the teaching evaluation role to the first large model, so that the first large model evaluates the courseware script from the perspectives of the learning subject role and the teaching evaluation role, respectively, to obtain first learning subject role evaluation information and first teaching evaluation role evaluation information output by the first large model; Sending the courseware script, the second prompt word of the learning subject role, and the second prompt word of the teaching evaluation role to the second large model, so that the second large model evaluates the courseware script from the perspectives of the learning subject role and the teaching evaluation role, respectively, to obtain second learning subject role evaluation information and second teaching evaluation role evaluation information output by the second large model; The comprehensive evaluation feedback information obtained based on the courseware evaluation information output by the at least two evaluation models includes: obtaining comprehensive evaluation feedback information based on the first learning subject role evaluation information, the first teaching evaluation role evaluation information, the second learning subject role evaluation information and the second teaching evaluation role evaluation information.

4. The method according to claim 3, characterized in that The comprehensive evaluation feedback information obtained based on the first learning subject role evaluation information, the first teaching evaluation role evaluation information, the second learning subject role evaluation information, and the second teaching evaluation role evaluation information includes: Combining the first learning subject role evaluation information and the second learning subject role evaluation information to obtain learning subject comprehensive information; Combining the first teaching evaluation role evaluation information and the second teaching evaluation role evaluation information to obtain comprehensive teaching evaluation information; Deduplication and clustering are performed on the learning subject comprehensive information and the teaching evaluation comprehensive information to obtain learning subject processing information and teaching evaluation processing information; The learning subject processing information and the teaching evaluation processing information are labeled with different evaluation dimensions and weights to obtain comprehensive evaluation feedback information.

5. According to the method of claim 4, the step of labeling the learning subject processing information and the teaching evaluation processing information with different evaluation dimensions and weights to obtain comprehensive evaluation feedback information comprises: Annotating the learning subject processing information and the teaching evaluation processing information with consensus evaluation dimensions and conflict evaluation dimensions to obtain learning subject annotation information and teaching evaluation annotation information; Assigning a first weight to the learning subject annotation information, and assigning a second weight to the teaching evaluation annotation information, wherein the first weight is greater than the second weight; Multiplying the learning subject annotation information by the first weight and adding the teaching evaluation annotation information by the second weight to obtain comprehensive annotation data; The comprehensive annotation data is evaluated to obtain comprehensive evaluation feedback information.

6. The method according to claim 1 or 2, characterized in that The process of generating the courseware script to be evaluated based on the teaching demand information of the education subject includes: Receive teaching demand information from educational subjects; Determining a teaching knowledge base based on the teaching demand information; The teaching demand information, the teaching knowledge base and the prompt words of the education subject are sent to the execution model to obtain the courseware script to be evaluated output by the execution model.

7. The method according to any one of claims 1 to 5, wherein: The courseware script is optimized based on the comprehensive evaluation feedback information to obtain the optimized courseware script including: Inputting the comprehensive evaluation feedback information into the execution model to obtain an analytical feedback report output by the execution model that includes actionable feedback items; Decomposing the parsing feedback report into at least one optimization task; Based on the courseware script, executing an operation corresponding to the at least one optimization task; In response to the completion of execution of the at least one optimization task, the executed script is checked to obtain an optimized courseware script.

8. A courseware script processing device, comprising: The script generation unit is configured to generate a courseware script to be evaluated based on the teaching demand information of the education subject; A prompt word generating unit is configured to generate role prompt words corresponding to different script evaluation roles for at least two evaluation large models based on the courseware script; a sending unit configured to send the courseware script and the role prompt word to the at least two evaluation large models, so that each evaluation large model evaluates the courseware script from the perspective of the script evaluation role corresponding to the role prompt word, and obtains courseware evaluation information output by the at least two evaluation large models; An evaluation unit configured to obtain comprehensive evaluation feedback information based on the courseware evaluation information output by the at least two evaluation models; The optimization unit is configured to optimize the courseware script based on the comprehensive evaluation feedback information to obtain an optimized courseware script.

9. An electronic device, characterized in that: include: at least one processor; and a memory communicatively coupled to the at least one processor; The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to enable the computer to execute the method according to any one of claims 1 to 7.

Images