A teaching data generation method, device, equipment and storage medium

Abstract

This invention provides a method, apparatus, device, and storage medium for generating teaching data. The method includes: acquiring user-inputted unit instructional design requirements; calling a pre-set domain-specific model; acquiring teaching intentions based on the unit instructional design requirements; retrieving knowledge information related to the teaching intentions from a domain knowledge base; calling the domain-specific model; and generating unit instructional design content based on the unit instructional design requirements and the knowledge information related to the teaching intentions. Furthermore, it can generate lesson-level teaching activity content based on the unit instructional design content, and can generate lesson-level teaching activity courseware based on the lesson-level teaching activity content. The teaching data generation method provided by this invention can intelligently generate teaching data, which can be directly used or modified by the user, and can also serve as a reference for the user's instructional data design, thereby greatly reducing the user's burden.

Description

Technical Field

[0001] This invention relates to the field of information generation, and more particularly to a method, apparatus, device, and storage medium for generating teaching data. Background Technology

[0002] Teaching data (such as instructional designs) is the data that teachers use to conduct their teaching. Currently, teaching data is usually designed by teachers based on curriculum standards, teaching content, and students. Designing teaching data is a difficult and arduous task. If teaching data could be generated intelligently, it would greatly reduce the burden on teachers. However, how to intelligently generate teaching data is a problem that urgently needs to be solved. Summary of the Invention

[0003] In view of this, the present invention provides a method, apparatus, device, and storage medium for generating teaching data, thereby reducing the burden on teachers. The technical solution is as follows:

[0004] Firstly, a method for generating teaching data is provided, including:

[0005] Unit instructional design requirements for obtaining user input;

[0006] Invoke the pre-set domain model and obtain the teaching intent according to the unit instructional design requirements;

[0007] Retrieve knowledge information related to the stated teaching intent from the domain knowledge base;

[0008] The domain-wide model is invoked, and unit instructional design content is generated based on the unit instructional design requirements and knowledge information related to the instructional intent, serving as the target unit instructional design content.

[0009] The domain-specific large model is obtained by training data from the education field and simultaneously training on specified tasks, which include at least an intent recognition task and a unit instructional design generation task.

[0010] Optionally, the training process of the domain-wide model includes:

[0011] The initial domain-wide model was trained using unlabeled training data from the education field to obtain the first domain-wide model;

[0012] In conjunction with the specified task, the first domain-specific large model is trained using labeled training data from the education domain to obtain the second domain-specific large model, which serves as the final domain-specific large model.

[0013] Optionally, the teaching data generation method further includes:

[0014] Obtain the first content to be modified selected by the user from the generated unit instructional design content, and obtain the user's modification requirements for the first content and the context information of the first content;

[0015] The domain model is invoked, and the first content is modified according to the modification requirements and context information of the first content. The modified unit instructional design content is used as the target unit instructional design content.

[0016] Optionally, the target unit instructional design content includes some or all of the following: unit theme, learning task group to which the unit belongs, unit teaching content, unit teaching objectives, situational task design content, and learning task design content.

[0017] The learning task design includes several task information items, each of which includes some or all of the following information: task theme, task objective, task activity information, and the number of class hours required for the task. The task activity information includes the name of the class hour teaching activity.

[0018] Optionally, the specified task may further include: a task for generating lesson-based teaching activities;

[0019] The teaching data generation method also includes:

[0020] Obtain the name of the lesson activity specified by the user from the instructional design content of the target unit, and use it as the name of the target lesson activity;

[0021] The domain-wide model is invoked, and the teaching activity content corresponding to the target lesson activity name is generated based on the target lesson activity name and the target unit instructional design content. This content is then used as the target lesson activity content.

[0022] Optionally, the teaching data generation method further includes:

[0023] Obtain the second content to be modified selected by the user from the generated lesson teaching activity content, and obtain the user's modification requirements for the second content and the context information of the second content;

[0024] The domain model is invoked, and the second content is modified according to the modification requirements and context information of the second content. The modified lesson teaching activity content is used as the target lesson teaching activity content.

[0025] Optionally, the teaching data generation method further includes:

[0026] Based on the content of the target lesson's teaching activities, generate lesson teaching activity courseware.

[0027] Optionally, the specified task may further include: a courseware outline generation task;

[0028] The step of generating lesson teaching activity courseware based on the target lesson teaching activity content includes:

[0029] The domain model is invoked to generate a courseware outline based on the target lesson teaching activity content. The courseware outline includes the key points of each page of the courseware.

[0030] Based on the courseware outline, generate courseware for each lesson's teaching activities.

[0031] Optionally, the specified task may further include: courseware content planning task;

[0032] The step of generating lesson teaching activity courseware based on the courseware outline includes:

[0033] The domain-wide model is invoked, and based on the key points of each page of the courseware, the constituent elements of each page of the courseware and the corresponding element content requirements are determined.

[0034] Based on the element content requirements corresponding to the constituent elements of each page of the courseware, obtain the element content corresponding to the constituent elements of each page of the courseware.

[0035] Based on the constituent elements of each page of the courseware and the content of the corresponding elements, generate courseware for each lesson's teaching activities.

[0036] Optionally, the specified task may further include: a keyword recognition task;

[0037] The step of obtaining the element content corresponding to the constituent elements of each page of the courseware according to the element content requirements of each page includes:

[0038] For each page of the presentation slides:

[0039] The domain-wide model is invoked to identify keywords from the content requirements corresponding to the constituent elements of the courseware on this page;

[0040] Based on the keywords, materials related to the keywords are retrieved from the multimodal material library and used as the element content corresponding to the constituent elements of the courseware on this page.

[0041] Optionally, the specified task may further include: material generation task;

[0042] The step of obtaining the element content corresponding to the constituent elements of each page of the courseware according to the element content requirements of each page of the courseware also includes:

[0043] If no material related to the keyword is found in the multimodal material library, the domain big model is invoked to generate the element content corresponding to the constituent elements of the courseware page according to the element content requirements of the constituent elements of the courseware page.

[0044] Optionally, the specified task may further include: a structured chart generation task and / or a keyword recognition task;

[0045] The teaching data generation method also includes:

[0046] Obtain the text content selected by the user from the generated lesson teaching activity courseware; call the domain big model to generate a structured chart based on the text content;

[0047] And / or,

[0048] Obtain the user's material acquisition needs; invoke the domain big model to identify keywords from the material acquisition needs; retrieve materials related to the identified keywords from the multimodal material library based on the identified keywords; add the retrieved materials to the generated lesson teaching activity courseware.

[0049] Secondly, a teaching data generation device is provided, including: a teaching design requirements acquisition module, a teaching intention acquisition module, a domain knowledge retrieval module, and a unit teaching design content generation module;

[0050] The instructional design requirements acquisition module is used to acquire the unit instructional design requirements input by the user.

[0051] The teaching intent acquisition module is used to call a pre-set domain model and acquire the teaching intent according to the unit teaching design requirements;

[0052] The domain knowledge retrieval module is used to retrieve knowledge information related to the teaching intention from the domain knowledge base;

[0053] The unit instruction design content generation module is used to call the domain big model and generate unit instruction design content based on the unit instruction design requirements and knowledge information related to the teaching intention, which serves as the target unit instruction design content.

[0054] The domain-specific large model is obtained by training data from the education field and simultaneously training on specified tasks, which include at least an intent recognition task and a unit instructional design generation task.

[0055] Thirdly, a teaching data generation device is provided, including: a memory and a processor;

[0056] The memory is used to store programs;

[0057] The processor is used to execute the program to implement each step of the teaching data generation method described above.

[0058] Fourthly, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the steps of the teaching data generation method described in any of the preceding claims.

[0059] The teaching data generation method provided by this invention first obtains the unit instructional design requirements input by the user, then calls a pre-set domain-wide model to obtain the teaching intent based on the unit instructional design requirements, next retrieves knowledge information related to the teaching intent from the domain knowledge base, and finally calls the domain-wide model to generate the unit instructional design content based on the unit instructional design requirements and the knowledge information related to the teaching intent. The teaching data generation method provided by this invention can intelligently generate unit instructional design content according to the user's unit instructional design requirements. The generated unit instructional design content can be directly used by the user or modified for use, and can also serve as a reference for the user in designing instructional data, thereby greatly reducing the user's burden. Attached Figure Description

[0060] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0061] Figure 1 A flowchart illustrating a teaching data generation method provided in an embodiment of the present invention;

[0062] Figure 2 An example of a unit instruction design interface provided in an embodiment of the present invention;

[0063] Figure 3 Another example of a unit instruction design interface provided in an embodiment of the present invention;

[0064] Figure 4 Examples of unit instruction design outlines and specific unit instruction design content provided for embodiments of the present invention;

[0065] Figure 5 A flowchart illustrating the process of obtaining teaching activity content for class periods, provided in an embodiment of the present invention;

[0066] Figure 6 This is a flowchart illustrating the process of generating courseware for a lesson based on the content of the lesson's teaching activities, as provided in an embodiment of the present invention.

[0067] Figure 7 This invention provides an example of converting text in courseware for lesson teaching activities into mind maps.

[0068] Figure 8 A schematic diagram of the structure of the teaching data generation device provided in an embodiment of the present invention;

[0069] Figure 9 A schematic diagram of the structure of the teaching data generation device provided in an embodiment of the present invention. Detailed Implementation

[0070] Before introducing the proposed solution, let's first explain the English terms used in this document:

[0071] Prompt: Instructions. When conversing with AI (such as large artificial intelligence models), you need to send instructions to the AI. These can be a text description or a parameter description in a certain format.

[0072] Large-scale artificial intelligence models, also known as large-scale deep learning models, are artificial intelligence models based on deep learning technology. They consist of hundreds of millions of parameters and can achieve complex tasks through learning and training on massive amounts of data. The domain-specific large model in this invention is a large-scale artificial intelligence model.

[0073] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0074] Please see Figure 1 The diagram illustrates a flowchart of a teaching data generation method provided by an embodiment of the present invention, which may include:

[0075] Step S101: Obtain the unit instruction design requirements input by the user.

[0076] Users can input unit instruction design requirements in the unit instruction design interface. Please refer to [link / reference]. Figure 2 This shows an example of a unit instructional design interface. Figure 2 The phrase "I want a teaching design for 'Quality of the Times, Craftsmanship Spirit'" is the unit teaching design requirement entered by the user.

[0077] In this embodiment, the unit instruction design requirements input by the user may include, but are not limited to, some or all of the following information: textbook information (e.g., compulsory Chinese textbook for senior high school students), unit information (e.g., which unit), the names of the texts included in the unit, the unit theme (referring to the teaching theme of this unit as defined in the textbook), the unit teaching objectives, and the unit teaching methods, etc.

[0078] For example, the user-input unit teaching design requirement can be "Generate a teaching design for the second unit of the compulsory second semester of senior high school Chinese, on the theme of contemporary quality and craftsmanship spirit, which should reflect the unit teaching philosophy." In this example, "compulsory second semester of senior high school Chinese" is the textbook information, "second unit" is the unit information, "contemporary quality and craftsmanship spirit" is the unit theme, and "reflect the unit teaching philosophy" is the unit teaching method.

[0079] To improve the effectiveness of instructional design generation, the unit instructional design interface can display input guidance information. This guidance information is used to direct users to input key information, thereby improving the quality of the generated instructional design content. Please refer to [link / reference]. Figure 3 This shows another example of a unit instructional design interface. Figure 3 The shown unit lesson design interface includes an input box with guidance information such as "Please fill in the unit theme, included text titles, unit learning objectives..." Users can enter the unit lesson design requirements in the input box according to the guidance information. Figure 3 The "Unit Theme: Dream of the Red Chamber", "Text Title: Reading the Whole Book Dream of the Red Chamber", and "Unit Learning Objectives: Read the whole book Dream of the Red Chamber and understand the author..." shown in the text are the unit teaching design requirements entered by the user based on the input guidance information.

[0080] Additionally, it should be noted that users can input unit lesson plan requirements via text input or voice input, such as... Figure 3 As shown, users can directly... Figure 3 Text input can be made in the input box, or voice input can be made by long-pressing the voice input icon in the lower right corner of the input box. If the unit teaching design requirement obtained in step S101 is that the user inputs text, then after obtaining the user input, the subsequent step S102 is executed directly. If the unit teaching design requirement obtained in step S101 is that the user inputs voice, then the voice is converted to text, and then the subsequent step S102 is executed.

[0081] Step S102: Call the pre-set domain model and obtain the teaching intent according to the unit teaching design requirements.

[0082] In this process, the pre-set domain model is invoked, and the teaching intentions obtained according to the unit teaching design requirements can be, but are not limited to, some or all of the textbook version, unit name, keywords, etc.

[0083] For example, if the unit teaching design requirement is "Please help me generate a large unit teaching design for reading the whole book 'Dream of the Red Chamber' in Unit 7 of the compulsory second semester of high school Chinese", then the pre-set domain model will be called. According to the unit teaching design requirement, the teaching intentions obtained can be the textbook version "compulsory second semester of high school Chinese", the unit name "Unit 7", the keyword "Dream of the Red Chamber", etc.

[0084] Specifically, the process of calling a pre-configured domain model to obtain teaching intentions based on unit instructional design requirements may include: obtaining a pre-configured first prompt format template, which includes unit instructional design requirement information slots, and is used to instruct the domain model to identify teaching intentions based on the information in the unit instructional design requirement information slots; filling the unit instructional design requirements input by the user into the unit instructional design requirement information slots to obtain an edited first instruction prompt; and inputting the edited first instruction prompt into the domain model to obtain the teaching intentions output by the domain model.

[0085] Step S103: Retrieve knowledge information related to the teaching intention from the domain knowledge base.

[0086] The domain knowledge base is a knowledge base for the education domain. It includes at least knowledge related to teaching scenarios, and may also include knowledge from other scenarios within the education domain. The knowledge related to teaching scenarios in the domain knowledge base may include several fields and the content corresponding to each field.

[0087] For example, the knowledge of teaching scenarios in the domain knowledge base may include fields such as unit introduction, unit theme, course name, and keywords, as well as the content corresponding to each field. Accordingly, the search results obtained by retrieving knowledge information related to teaching intentions from the domain knowledge base are the content information corresponding to the fields of unit introduction, unit theme, course name, and keywords.

[0088] Each piece of knowledge in the domain knowledge base has a vector representation. When retrieving knowledge information related to teaching intentions from the domain knowledge base, the vector representation of the teaching intentions can be obtained first. Then, the similarity between the vector representation of the teaching intentions and the vector representation of each piece of knowledge in the teaching scenario in the domain knowledge base can be calculated. Finally, the knowledge information related to the teaching intentions can be determined based on the calculated similarity.

[0089] Step S104: Call the domain-wide model and generate unit instruction design content based on the unit instruction design requirements and knowledge information related to the teaching intent.

[0090] The domain-wide model generates unit instructional design content based on the unit instructional design requirements and background knowledge related to the teaching intent.

[0091] Optionally, the generated unit instructional design content may include a unit instructional design outline and specific unit instructional design content. Please refer to [link / reference]. Figure 4 It shows examples of unit instructional design outlines and specific unit instructional design content. Figure 4 The left side of the text contains the unit teaching design outline, and the right side contains the specific unit teaching design content.

[0092] The process of generating unit instruction design content by invoking the domain-wide model, based on unit instruction design requirements and knowledge information related to teaching intentions, may include: obtaining a pre-configured second prompt format template, which includes unit instruction design requirement information slots and domain knowledge information slots. This template instructs the domain-wide model to generate unit instruction design content based on the information in the unit instruction design requirement information slots and the information in the domain knowledge information slots; filling the unit instruction design requirements into the unit instruction design requirement information slots and the knowledge information related to teaching intentions into the domain knowledge information slots, resulting in an edited second instruction prompt; inputting the edited second instruction prompt into the domain-wide model to obtain the structured content of the unit instruction design output by the model; parsing the structured content of the unit instruction design after obtaining it, and then calling the instruction design plugin to generate the final unit instruction design content that can be presented to the user.

[0093] Optionally, the unit instruction design may include some or all of the following: unit theme, the learning task group to which the unit belongs (the learning task group refers to the sixteen major Chinese language learning abilities defined in the new curriculum standards), unit teaching content, unit teaching objectives, situational task design content, and learning task design content. Among them, the learning task design content includes several task information, and each task information may include some or all of the following: task theme, task objective, task activity information, and the number of class hours required for the task. The task activity information includes the name of the teaching activity for each class hour.

[0094] The following is an example of the unit instructional design content generated using steps S101 to S104 above:

[0095] Table 1 Example of Unit Teaching Design Content

[0096]

[0097]

[0098]

[0099] Optionally, after generating the unit instructional design content, the system can obtain the first piece of content selected by the user from the generated content, along with the user's modification requirements and contextual information. Then, it can invoke the domain-wide model to modify the first piece of content based on these requirements and contextual information, resulting in the modified unit instructional design content. In other words, after generating the unit instructional design content, the user can select a portion of the content and input modification requirements, triggering modifications to the selected content. Alternatively, the user can also directly and independently modify the generated unit instructional design content.

[0100] In this embodiment, the domain-wide model is obtained by training data from the education domain (including at least training data from teaching scenarios) and simultaneously training it with a specified task. Specifically, firstly, the initial domain-wide model is trained using unlabeled training data from the education domain (i.e., unsupervised training) to enable the domain-wide model to learn relevant knowledge in the education domain, resulting in a first domain-wide model. Then, the first domain-wide model is trained using labeled training data from the education domain (i.e., supervised training) to enable the domain model to perform the specified task, resulting in a second domain-wide model, which serves as the final domain-wide model.

[0101] The specified tasks include at least an intent recognition task and a unit instructional design generation task. It should be noted that training in conjunction with the intent recognition task is to enable the domain model to acquire the teaching intent according to the unit instructional design requirements, and training in conjunction with the unit instructional design generation task is to enable the domain model to generate unit instructional design content.

[0102] Optionally, during unsupervised training, parts of the training data can be masked or replaced, allowing the domain-wide model to predict the masked or replaced data. During supervised training, labeled training data corresponding to a specified task can be used. For example, the training data for the intent recognition task includes unit instruction design requirements and prompts to instruct the domain-wide model to execute the intent recognition task. The labeled information is the standard instruction intent. During training, the training data for the intent recognition task is input into the domain-wide model, with the goal of training the model to obtain instruction intents that approximate the standard instruction intents. Similarly, the training data for the unit instruction design task includes unit instruction design requirements and prompts to instruct the domain-wide model to execute the unit instruction design task. During training, the training data for the unit instruction design task is input into the domain-wide model, with the goal of training the unit instruction design content generated by the domain-wide model to approximate the standard unit instruction design content.

[0103] The teaching data generation method provided in this invention first obtains the unit instructional design requirements input by the user, then calls a pre-set domain-wide model to obtain the teaching intent based on the unit instructional design requirements, next retrieves knowledge information related to the teaching intent from the domain knowledge base, and finally calls the domain-wide model to generate unit instructional design content based on the unit instructional design requirements and the knowledge information related to the teaching intent. The teaching data generation method provided in this invention can intelligently generate unit instructional design content according to the user's unit instructional design requirements. The generated unit instructional design content can be directly used by the user or modified for use, and can also serve as a reference for the user in designing instructional data, thereby greatly reducing the user's burden.

[0104] This invention provides another method for generating teaching data. This method differs from the method provided in the above embodiments in that, in addition to the process of obtaining the unit instructional design content (i.e., the steps provided in the above embodiments), it also includes a process of obtaining the content of lesson-level teaching activities. Please refer to [link to relevant documentation]. Figure 5 The diagram illustrates the process of obtaining the content of a lesson's teaching activities, which may include:

[0105] Step S501: Obtain the name of the lesson activity specified by the user from the unit instructional design content, and use it as the name of the target lesson activity.

[0106] The unit instruction design content can be the unit instruction design content obtained by using steps S101 to S105 in the above embodiments, or it can be the unit instruction design content after modification of the unit instruction design content obtained by using steps S101 to S105 in the above embodiments.

[0107] The above embodiments mention that the unit instructional design content may include learning task design content, which may include task activity information. The task activity information may include the name of the teaching activity for each lesson. Users can select the name of the teaching activity for each lesson from the unit instructional design content, thereby triggering the generation of the teaching activity content for each lesson.

[0108] Step S502: Call the preset domain model and generate the teaching activity content corresponding to the target lesson activity name based on the target lesson activity name and the target unit teaching design content.

[0109] Specifically, the process of calling a pre-configured domain model to generate the content of a lesson activity corresponding to the target lesson activity name and the unit teaching design content can include: obtaining a pre-configured third prompt format template, which includes an activity name information slot and a unit teaching design information slot. The third prompt format template is used to instruct the domain model to generate the lesson activity content based on the information in the activity name information slot and the information in the unit teaching design information slot; filling the target lesson activity name into the activity name information slot and the unit teaching design content into the unit teaching design information slot to obtain an edited third instruction prompt; inputting the edited third instruction prompt into the domain model to obtain the structured content of the lesson activity output by the domain model; after obtaining the structured content of the lesson activity, the structured content of the lesson activity can be parsed, and then the lesson activity plugin can be called to generate the final lesson activity content that can be presented to the user.

[0110] Optionally, after selecting a lesson activity name from the unit instructional design content, the user can add some requirements, such as requirements for the activity theme, student learning situation, and teaching habits. The generation of the lesson activity content is then triggered after the user selects the lesson activity name and enters the requirements. It should be noted that if the user enters some requirements, a pre-defined domain model is invoked. Based on the target lesson activity name, the unit instructional design content, and the user's input requirements, the lesson activity content corresponding to the target lesson activity name is generated (the third prompt format template mentioned above also includes user requirement information slots, which then need to be filled with the user's input requirements).

[0111] For example, the content of a lesson activity may include one or more of the following: activity introduction, activity objectives, activity process, activity summary, activity extension, and homework assignment. Table 2 below provides an example of the content of a lesson activity generated based on the above method. It is the lesson activity content corresponding to "Activity 1: Identifying the Content of *Dream of the Red Chamber* - Chapter Titles" in "Task 1: Understanding the Story of *Dream of the Red Chamber*" in Table 1 above.

[0112] Table 2 Examples of Class Hour Teaching Activities

[0113]

[0114] Optionally, after generating the lesson activity content corresponding to the target lesson activity name, the system can obtain the second content to be modified selected by the user from the generated lesson activity content, along with the user's modification requirements and contextual information of the second content. Then, it can call the domain-wide model to modify the second content based on the modification requirements and contextual information. In other words, after generating the lesson activity content, the user can select a portion of the content and input modification requirements, thus triggering modifications to the selected content. Alternatively, the user can directly modify the generated lesson activity content.

[0115] It should be noted that the domain-wide model in this embodiment also uses training data from the education domain (including at least training data from teaching scenarios) and is trained using specified tasks (first unsupervised training, then supervised training using the specified tasks). These specified tasks include not only intent recognition and unit instruction design generation tasks, but also lesson-based teaching activity generation tasks. Training with lesson-based teaching activity generation tasks is to enable the domain-wide model to generate lesson-based teaching activity content.

[0116] The training data corresponding to the task of generating teaching activities for each lesson consists of a prompt containing the name of the teaching activity and the content of the unit teaching design, which is used to instruct the domain model to execute the task of generating teaching activities for each lesson. The annotation information is the standard teaching activity content corresponding to the name of the teaching activity in the prompt. During training, the training data corresponding to the task of generating teaching activities for each lesson is input into the domain model, with the goal of training the teaching activities generated based on the domain model to be close to the standard teaching activities.

[0117] The teaching data generation method provided in this invention first generates unit teaching design content based on the user-input unit teaching design requirements. After obtaining the unit teaching design content, it can generate lesson teaching activity content corresponding to the lesson teaching activity names specified by the user from the unit teaching design content. This method not only intelligently generates unit teaching design content based on the user's unit teaching design requirements, but also intelligently generates corresponding lesson teaching activity content for the lesson teaching activity names within the unit teaching design content. The generated unit teaching design content and lesson teaching activity content can be directly used or modified by the user, and can also serve as a reference for the user's teaching data design, thereby greatly reducing the user's burden.

[0118] This invention provides another teaching data generation method, which differs from the teaching data generation method provided in the second embodiment above in that, in addition to the process of obtaining unit teaching design content (the step of obtaining unit teaching design content in the first embodiment) and the process of obtaining lesson teaching activity content (the step of obtaining lesson teaching activity content in the second embodiment), it also includes: generating lesson teaching activity courseware based on the obtained lesson teaching activity content.

[0119] Please see Figure 6 The diagram illustrates the process of generating lesson plan materials based on the content of the lesson activities, which may include:

[0120] Step S601: Call the domain big model and generate a courseware outline based on the teaching activities of the target class period.

[0121] The lesson plan outline includes the key points of each page of the lesson plan.

[0122] Specifically, the process of calling the domain-wide model to generate a courseware outline based on the target lesson's teaching activities can include: obtaining a pre-configured fourth prompt format template, which includes lesson teaching activity information slots and is used to instruct the domain-wide model to generate a courseware outline based on the information in the lesson teaching activity information slots; filling the target lesson teaching activity content into the lesson teaching activity information slots to obtain an edited fourth prompt instruction; and inputting the edited fourth prompt instruction into the domain-wide model to obtain the courseware outline output by the domain-wide model.

[0123] Step S602: Generate lesson teaching activity courseware based on the courseware outline.

[0124] Specifically, based on the courseware outline, the process of generating lesson-specific teaching activity courseware may include:

[0125] Step S6021: Call the domain big model and determine the constituent elements of each page of the courseware and the corresponding element content requirements based on the key points of each page of the courseware.

[0126] It should be noted that the constituent elements of a single slide indicate the type of information required for that slide. For example, the constituent elements of a single slide can be text and images; that is, the constituent elements of a single slide can be one or more combinations of text, images, video, and audio. The element content requirements indicate the required element content. For example, the element content requirements for text indicate what kind of text is needed, for images indicate what kind of images are needed, for videos indicate what kind of video content is needed, and for audio content, they indicate what kind of audio content is needed.

[0127] Step S6022: Based on the element content requirements corresponding to the constituent elements of each page of the courseware, obtain the element content corresponding to the constituent elements of each page of the courseware.

[0128] Optionally, for each courseware, the domain big model is first invoked to identify keywords from the element content requirements corresponding to the constituent elements of the courseware on that page. Then, based on the identified keywords, materials related to the keywords are retrieved from the multimodal material library and used as the element content corresponding to the constituent elements of the courseware on that page.

[0129] The multimodal resource library includes materials of various modalities, such as text, images, videos, and audio. Assuming a single slide consists of text and images, keywords are identified from the content requirements corresponding to the text elements. Based on these keywords, related text materials are retrieved from the multimodal resource library. Similarly, keywords are identified from the content requirements corresponding to the images, and related image materials are retrieved from the multimodal resource library.

[0130] The process of calling the domain-wide model to identify keywords from the content requirements of the constituent elements of the courseware page may include: obtaining a pre-configured fifth prompt template, which includes requirement information slots and is used to instruct the domain-wide model to identify keywords from the information in the requirement information slots; filling the content requirements of the constituent elements of the courseware page into the requirement information slots in the fifth prompt template to obtain the edited fifth prompt; and inputting the edited fifth prompt into the domain-wide model to obtain the keywords output by the domain-wide model.

[0131] In some cases, it may be possible to find no relevant materials in the multimodal material library based on identified keywords. In such situations, a domain-wide model can be invoked to generate element content according to the element content requirements. For example, a single slide may consist of text and an image. If no relevant text or image is found in the multimodal material library based on keywords identified from the element content requirements corresponding to the text, then the domain-wide model can be invoked to generate the text and the image, respectively.

[0132] Of course, after obtaining the element content requirements corresponding to the constituent elements of each page of courseware through step S6021, the domain big model can also be directly called to generate the element content corresponding to the constituent elements of each page of courseware according to the element content requirements corresponding to the constituent elements of each page of courseware.

[0133] Step S6023: Generate lesson teaching activity courseware based on the constituent elements of each page of the courseware and the corresponding element content of each page of the courseware.

[0134] Specifically, based on the constituent elements of each page of the courseware and the corresponding content of each constituent element, a courseware module (such as a PPT template) is selected from the courseware template library. Then, based on the content of each page's constituent elements and the selected courseware template, a lesson-specific teaching activity courseware is generated. The process of generating the lesson-specific teaching activity courseware based on the content of each page's constituent elements and the selected courseware template includes: filling the content of each page's constituent elements into the selected courseware template and then formatting and beautifying it.

[0135] Optionally, after generating the lesson activity courseware, users can modify it. When modifying the courseware, users can independently edit it or select content to be modified, triggering changes to the selected content. For the selected content, the domain-wide model can be invoked to modify it. For example, if a user finds a piece of text too long and wants to convert it into a structured diagram (such as a mind map), the user can select that text. The domain-wide model can then be invoked to generate a structured diagram (such as a mind map) based on the selected text. Please refer to [link to relevant documentation]. Figure 7 This example demonstrates how to convert text in lesson plan materials into mind maps.

[0136] Optionally, after generating the courseware for the lesson activities, in addition to calling the domain big model to modify the content selected by the user, it can also obtain the material acquisition requirements input by the user, call the domain big model to identify keywords from the material acquisition requirements, retrieve materials from the multimodal material library based on the identified keywords, and then add the retrieved materials to the courseware for the lesson activities.

[0137] For example, a user can input "Please provide me with some pictures depicting the gate of the Jia family mansion from the text 'Dream of the Red Chamber'". After receiving the user's material acquisition request, the domain model can be invoked to identify keywords from the request. Assuming the keywords "Dream of the Red Chamber" and "Jia family mansion gate" are identified, the model further retrieves image materials related to "Dream of the Red Chamber" and "Jia family mansion gate" from the multimodal material library, and then adds the retrieved image materials to the lesson's teaching activity materials. If no image materials related to "Dream of the Red Chamber" and "Jia family mansion gate" are found in the multimodal material library, the domain model can be invoked to generate image materials related to "Dream of the Red Chamber" and "Jia family mansion gate".

[0138] It should be noted that the domain-specific large model in this embodiment also uses training data from the education domain (including at least training data from teaching scenarios) and is trained using specified tasks (first unsupervised training, then supervised training using specified tasks). These specified tasks include not only intent recognition tasks, unit instruction design generation tasks, and lesson activity generation tasks, but also courseware outline generation tasks, courseware content planning tasks, keyword recognition tasks, material generation tasks (such as text generation tasks, image generation tasks, audio generation tasks, video generation tasks, etc.), and structured chart generation tasks (such as mind map generation tasks, table generation tasks, etc.).

[0139] The training methods for all the above tasks are the same, only the training data differs. For example, the training data for the courseware outline generation task includes the content of the teaching activities for each lesson and is used to instruct the domain-wide model to execute the courseware outline generation task. The annotation information is the standard courseware outline. The training data for the courseware content planning task includes the courseware outline and is used to instruct the domain-wide model to execute the courseware content planning task. The annotation information is the standard courseware content planning information. The training data for the keyword recognition task includes the material content requirements and is used to instruct the domain-wide model to execute the keyword recognition task. The annotation information is the real keywords in the material content requirements. The training data for the material generation task includes the material content requirements and is used to instruct the domain-wide model to execute the material generation task. The annotation information is the standard data corresponding to the material content requirements. The training data for the structured chart generation task includes a piece of text and is used to instruct the domain-wide model to execute the structured chart generation task. The annotation information is the standard structured chart corresponding to the text.

[0140] The teaching data generation method provided in this invention first generates unit teaching design content based on the user-input unit teaching design requirements. After obtaining the unit teaching design content, it can generate lesson teaching activity content corresponding to the lesson teaching activity names specified by the user from the unit teaching design content. After obtaining the lesson teaching activity content, it can generate lesson teaching activity courseware based on the lesson teaching activity content. This teaching data generation method not only intelligently generates unit teaching design content based on the user's unit teaching design requirements, but also intelligently generates corresponding lesson teaching activity content for the lesson teaching activity names in the unit teaching design content, and intelligently generates lesson teaching activity courseware based on the lesson teaching activity content. This generated teaching data can be directly used or modified by the user, and can also serve as a reference for the user's teaching data design, thereby greatly reducing the user's burden.

[0141] This invention provides a teaching data generation device. The teaching data generation device provided in this invention is described below. The teaching data generation device described below can be referred to in correspondence with the teaching data generation method described above.

[0142] Please see Figure 8 The diagram shows a schematic of the teaching data generation device provided in an embodiment of the present invention. The teaching data generation device may include: a teaching design requirements acquisition module 801, a teaching intention acquisition module 802, a domain knowledge retrieval module 803, and a unit teaching design content generation module 804.

[0143] The instructional design requirements acquisition module 801 is used to acquire the unit instructional design requirements input by the user.

[0144] The teaching intent acquisition module 802 is used to call a pre-set domain model and acquire teaching intent according to the unit teaching design requirements.

[0145] The domain knowledge retrieval module 803 is used to retrieve knowledge information related to the teaching intent from the domain knowledge base.

[0146] The unit instructional design content generation module 804 is used to call the domain-wide model and generate unit instructional design content based on the unit instructional design requirements and knowledge information related to the teaching intention, which serves as the target unit instructional design content.

[0147] The domain-specific large model is obtained by training data from the education field and simultaneously training on specified tasks, which include at least an intent recognition task and a unit instructional design generation task.

[0148] Optionally, the teaching data generation device provided in this embodiment of the invention may further include: a model training module. The model training module is used for:

[0149] The initial domain-wide model was trained using unlabeled training data from the education field to obtain the first domain-wide model;

[0150] In conjunction with the specified task, the first domain-specific large model is trained using labeled training data from the education domain to obtain the second domain-specific large model, which serves as the final domain-specific large model.

[0151] Optionally, the teaching data generation device provided in this embodiment of the invention may further include: a first content acquisition module, a first modification requirement acquisition module, a first context information acquisition module, and a first modification module.

[0152] The first content acquisition module is used to acquire the first content to be modified selected by the user from the generated unit instructional design content.

[0153] The first modification request acquisition module is used to acquire the user's modification request for the first content.

[0154] The first context information acquisition module is used to acquire the context information of the first content.

[0155] The first modification module is used to call the domain big model, modify the first content according to the modification requirements of the first content and the context information of the first content, and the modified unit teaching design content is used as the target unit teaching design content.

[0156] Optionally, the target unit instructional design content includes some or all of the following: unit theme, learning task group to which the unit belongs, unit teaching content, unit teaching objectives, situational task design content, and learning task design content.

[0157] The learning task design includes several task information items, each of which includes some or all of the following information: task theme, task objective, task activity information, and the number of class hours required for the task. The task activity information includes the name of the class hour teaching activity.

[0158] Optionally, the specified task may also include: a task for generating teaching activities for class periods.

[0159] The teaching data generation device may also include: a module for generating teaching activity content for each lesson.

[0160] The lesson activity content generation module is used for:

[0161] Obtain the name of the lesson activity specified by the user from the instructional design content of the target unit, and use it as the name of the target lesson activity;

[0162] The domain-wide model is invoked, and the teaching activity content corresponding to the target lesson activity name is generated based on the target lesson activity name and the target unit instructional design content. This content is then used as the target lesson activity content.

[0163] Optionally, the teaching data generation device may further include: a second content acquisition module, a second modification requirement acquisition module, a second context information acquisition module, and a second modification module.

[0164] The second content acquisition module is used to acquire the second content to be modified selected by the user from the generated lesson teaching activity content.

[0165] The second modification request acquisition module is used to acquire the user's modification request for the second content.

[0166] The second context information acquisition module is used to acquire the context information of the second content;

[0167] The second modification module is used to call the domain big model, modify the second content according to the modification requirements of the second content and the context information of the second content, and the modified lesson teaching activity content is used as the target lesson teaching activity content.

[0168] Optionally, the teaching data generation device may also include: a module for generating courseware for lesson teaching activities.

[0169] The lesson teaching activity courseware generation module is used to generate lesson teaching activity courseware based on the target lesson teaching activity content.

[0170] Optionally, the specified task may further include: a courseware outline generation task;

[0171] The lesson activity courseware generation module, when generating lesson activity courseware based on the target lesson activity content, is specifically used for:

[0172] The domain model is invoked to generate a courseware outline based on the target lesson teaching activity content. The courseware outline includes the key points of each page of the courseware.

[0173] Based on the courseware outline, generate courseware for each lesson's teaching activities.

[0174] Optionally, the specified task may further include: courseware content planning task;

[0175] The lesson activity courseware generation module, when generating lesson activity courseware based on the aforementioned courseware outline, is specifically used for:

[0176] The domain-wide model is invoked, and based on the key points of each page of the courseware, the constituent elements of each page of the courseware and the corresponding element content requirements are determined.

[0177] Based on the element content requirements corresponding to the constituent elements of each page of the courseware, obtain the element content corresponding to the constituent elements of each page of the courseware.

[0178] Based on the constituent elements of each page of the courseware and the content of the corresponding elements, generate courseware for each lesson's teaching activities.

[0179] Optionally, the specified task may further include: a keyword recognition task;

[0180] The lesson plan generation module, when retrieving the element content corresponding to the constituent elements of each page of the lesson plan based on the element content requirements of each page, is specifically used for:

[0181] For each page of the presentation slides:

[0182] The domain-wide model is invoked to identify keywords from the content requirements corresponding to the constituent elements of the courseware on this page;

[0183] Based on the keywords, materials related to the keywords are retrieved from the multimodal material library and used as the element content corresponding to the constituent elements of the courseware on this page.

[0184] Optionally, the specified task may further include: material generation task;

[0185] The module for generating courseware for lesson teaching activities is also used for:

[0186] If no material related to the keyword is found in the multimodal material library, the domain big model is invoked to generate the element content corresponding to the constituent elements of the courseware page according to the element content requirements of the constituent elements of the courseware page.

[0187] Optionally, the specified task may further include: a structured chart generation task and / or a keyword recognition task;

[0188] The teaching data generation device may also include: a structured chart generation module and / or a material acquisition module.

[0189] The structured chart generation module is used to obtain the text content selected by the user from the generated lesson teaching activity courseware, call the domain big model, and generate structured charts based on the text content.

[0190] The material acquisition module is used to acquire the user's material acquisition needs, call the domain big model, identify keywords from the material acquisition needs, retrieve materials related to the identified keywords from the multimodal material library, and add the retrieved materials to the generated lesson teaching activity courseware.

[0191] The teaching data generation device provided in this embodiment of the invention can generate unit teaching design content based on user-input unit teaching design requirements. After obtaining the unit teaching design content, it can generate lesson teaching activity content corresponding to the lesson teaching activity names specified by the user from the unit teaching design content. After generating the lesson teaching activity content, it can generate lesson teaching activity courseware based on the lesson teaching activity content. The teaching data generation device provided in this embodiment of the invention can intelligently generate unit teaching design content, lesson teaching activity content, and lesson teaching activity courseware. This generated teaching data can be directly used by the user or used after modification, and can also be used as a reference for the user to design teaching data, thereby greatly reducing the user's burden.

[0192] This invention provides a teaching data generation device; please refer to [link / reference]. Figure 9 The diagram shows the structure of the teaching data generation device, which may include: a processor 901, a communication interface 902, a memory 903, and a communication bus 904.

[0193] In this embodiment of the invention, the number of processor 901, communication interface 902, memory 903, and communication bus 904 is at least one, and processor 901, communication interface 902, and memory 903 communicate with each other through communication bus 904.

[0194] The processor 901 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention.

[0195] The memory 903 may include high-speed RAM, or it may also include non-volatile memory, such as at least one disk storage device;

[0196] The memory stores a program, which the processor can call. The program is used for:

[0197] Unit instructional design requirements for obtaining user input;

[0198] Invoke the pre-set domain model and obtain the teaching intent according to the unit instructional design requirements;

[0199] Retrieve knowledge information related to the stated teaching intent from the domain knowledge base;

[0200] The domain-wide model is invoked, and unit instructional design content is generated based on the unit instructional design requirements and knowledge information related to the instructional intent, serving as the target unit instructional design content.

[0201] The domain-specific large model is obtained by training data from the education field and simultaneously training on specified tasks, which include at least an intent recognition task and a unit instructional design generation task.

[0202] Optionally, the refined and extended functions of the program can be found in the description above.

[0203] This invention also provides a readable storage medium that stores a program suitable for execution by a processor, the program being used for:

[0204] Unit instructional design requirements for obtaining user input;

[0205] Invoke the pre-set domain model and obtain the teaching intent according to the unit instructional design requirements;

[0206] Retrieve knowledge information related to the stated teaching intent from the domain knowledge base;

[0207] The domain-wide model is invoked, and unit instructional design content is generated based on the unit instructional design requirements and knowledge information related to the instructional intent, serving as the target unit instructional design content.

[0208] The domain-specific large model is obtained by training data from the education field and simultaneously training on specified tasks, which include at least an intent recognition task and a unit instructional design generation task.

[0209] Optionally, the refined and extended functions of the program can be found in the description above.

[0210] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0211] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0212] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for generating teaching data, characterized in that, include: Unit instructional design requirements for obtaining user input; Invoke the pre-set domain model and obtain the teaching intent according to the unit instructional design requirements; Retrieve knowledge information related to the stated teaching intent from the domain knowledge base; Obtain a pre-configured second prompt format template, which includes unit instruction design requirement information slots and domain knowledge information slots. The second prompt format template is used to instruct the domain big model to generate unit instruction design content based on the information in the unit instruction design requirement information slots and in combination with the information in the domain knowledge information slots. The unit instruction design requirements are filled into the unit instruction design requirements information slot, and the knowledge information related to the teaching intention is filled into the domain knowledge information slot, resulting in the edited second instruction prompt; Based on the second instruction prompt, the domain big model is invoked to generate unit instruction design content, which serves as the target unit instruction design content; The domain-specific large model is obtained by training data from the education field and simultaneously training on specified tasks, which include at least an intent recognition task and a unit instructional design generation task.

2. The teaching data generation method according to claim 1, characterized in that, The training process of the aforementioned domain-specific large model includes: The initial domain-wide model was trained using unlabeled training data from the education field to obtain the first domain-wide model; In conjunction with the specified task, the first domain-specific large model is trained using labeled training data from the education domain to obtain the second domain-specific large model, which serves as the final domain-specific large model.

3. The teaching data generation method according to claim 1, characterized in that, Also includes: Obtain the first content to be modified selected by the user from the generated unit instructional design content, and obtain the user's modification requirements for the first content and the context information of the first content; The domain model is invoked, and the first content is modified according to the modification requirements and context information of the first content. The modified unit instructional design content is used as the target unit instructional design content.

4. The teaching data generation method according to any one of claims 1 to 3, characterized in that, The target unit instructional design content includes some or all of the following: unit theme, learning task group to which the unit belongs, unit teaching content, unit teaching objectives, situational task design content, and learning task design content; The learning task design includes several task information items, each of which includes some or all of the following information: task theme, task objective, task activity information, and the number of class hours required for the task. The task activity information includes the name of the class hour teaching activity.

5. The teaching data generation method according to claim 4, characterized in that, The specified tasks also include: tasks for generating class-time teaching activities; The teaching data generation method also includes: Obtain the name of the lesson activity specified by the user from the instructional design content of the target unit, and use it as the name of the target lesson activity; The domain-wide model is invoked, and the teaching activity content corresponding to the target lesson activity name is generated based on the target lesson activity name and the target unit instructional design content. This content is then used as the target lesson activity content.

6. The teaching data generation method according to claim 5, characterized in that, Also includes: Obtain the second content to be modified selected by the user from the generated lesson teaching activity content, and obtain the user's modification requirements for the second content and the context information of the second content; The domain model is invoked, and the second content is modified according to the modification requirements and context information of the second content. The modified lesson teaching activity content is used as the target lesson teaching activity content.

7. The teaching data generation method according to claim 5, characterized in that, Also includes: Based on the content of the target lesson's teaching activities, generate lesson teaching activity courseware.

8. The teaching data generation method according to claim 7, characterized in that, The specified tasks also include: courseware outline generation task; The step of generating lesson teaching activity courseware based on the target lesson teaching activity content includes: The domain model is invoked to generate a courseware outline based on the target lesson teaching activity content. The courseware outline includes the key points of each page of the courseware. Based on the courseware outline, generate courseware for each lesson's teaching activities.

9. The teaching data generation method according to claim 8, characterized in that, The designated task also includes: courseware content planning task; The step of generating lesson teaching activity courseware based on the courseware outline includes: The domain-wide model is invoked, and based on the key points of each page of the courseware, the constituent elements of each page of the courseware and the corresponding element content requirements are determined. Based on the element content requirements corresponding to the constituent elements of each page of the courseware, obtain the element content corresponding to the constituent elements of each page of the courseware. Based on the constituent elements of each page of the courseware and the content of the corresponding elements, generate courseware for each lesson's teaching activities.

10. The teaching data generation method according to claim 9, characterized in that, The specified task also includes: keyword recognition task; The step of obtaining the element content corresponding to the constituent elements of each page of the courseware according to the element content requirements of each page includes: For each page of the presentation slides: The domain-wide model is invoked to identify keywords from the content requirements corresponding to the constituent elements of the courseware on this page; Based on the keywords, materials related to the keywords are retrieved from the multimodal material library and used as the element content corresponding to the constituent elements of the courseware on this page.

11. The teaching data generation method according to claim 10, characterized in that, The specified task also includes: material generation task; The step of obtaining the element content corresponding to the constituent elements of each page of the courseware according to the element content requirements of each page of the courseware also includes: If no material related to the keyword is found in the multimodal material library, the domain big model is invoked to generate the element content corresponding to the constituent elements of the courseware page according to the element content requirements of the constituent elements of the courseware page.

12. The teaching data generation method according to claim 8, characterized in that, The specified tasks also include: structured chart generation task and / or keyword recognition task; The teaching data generation method also includes: Obtain the text content selected by the user from the generated lesson teaching activity courseware; call the domain big model to generate a structured chart based on the text content; And / or, Obtain the user's material acquisition needs; invoke the domain big model to identify keywords from the material acquisition needs; retrieve materials related to the identified keywords from the multimodal material library based on the identified keywords; add the retrieved materials to the generated lesson teaching activity courseware.

13. A teaching data generation device, characterized in that, include: The module includes modules for obtaining instructional design requirements, obtaining instructional intentions, retrieving domain knowledge, and generating unit instructional design content. The instructional design requirements acquisition module is used to acquire the unit instructional design requirements input by the user. The teaching intent acquisition module is used to call a pre-set domain model and acquire the teaching intent according to the unit teaching design requirements; The domain knowledge retrieval module is used to retrieve knowledge information related to the teaching intention from the domain knowledge base; The unit instruction design content generation module is used to obtain a pre-configured second prompt format template. The second prompt format template includes unit instruction design requirement information slots and domain knowledge information slots. The second prompt format template is used to instruct the domain big model to generate unit instruction design content based on the information in the unit instruction design requirement information slots and the information in the domain knowledge information slots. The unit instruction design requirements are filled into the unit instruction design requirement information slots, and the knowledge information related to the teaching intentions is filled into the domain knowledge information slots to obtain an edited second instruction prompt. Based on the second instruction prompt, the domain big model is called to generate unit instruction design content, which serves as the target unit instruction design content. The domain-specific large model is obtained by training data from the education field and simultaneously training on specified tasks, which include at least an intent recognition task and a unit instructional design generation task.

14. A teaching data generation device, characterized in that, include: Memory and processor; The memory is used to store programs; The processor is used to execute the program to implement each step of the teaching data generation method as described in any one of claims 1 to 12.

15. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements each step of the teaching data generation method as described in any one of claims 1 to 12.

Images