Courseware generation method, electronic equipment and computer storage medium

Abstract

The invention discloses a courseware generation method, electronic equipment and a computer storage medium. The method comprises the following steps: receiving target creation information input by a user on a courseware creation page; obtaining an initial creation intention corresponding to the target creation information based on a large language model and a private domain knowledge base; the private domain knowledge base is a knowledge base constructed after preprocessing knowledge texts of a preset field; displaying an intention editing page of the initial creation intention; after a target creation intention edited by the user based on the initial creation intention is obtained, an initial creation outline corresponding to the target creation intention is obtained based on the large language model, the private domain knowledge base and the interactive tool library; displaying an outline editing page of the initial creation outline; after obtaining a target creation outline edited by the user based on the initial creation outline, generating courseware corresponding to the target creation outline based on the large language model, the private domain knowledge base, the interactive tool library and the multi-modal material library; and displaying a courseware display page including the courseware.

Description

Courseware generation method, electronic device and computer storage medium TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and particularly relates to a courseware generation method, an electronic device and a computer storage medium. BACKGROUND

[0002] With the continuous development of artificial intelligence (AI) technology, intelligent office products have gradually become popular. At present, generative AI has shown great potential and advantages in content creation and editing, information search and extraction, etc. For example, some office software integrated with large language models (LLM) can assist users in creating courseware, greatly improving the user's office efficiency.

[0003] However, the generation of courseware in the prior art only relies on large language models, lacks pertinence, and the generated results have certain randomness and may contain inaccurate information, which cannot accurately meet the actual teaching needs.

[0004] SUMMARY

[0005] The embodiments of the present application provide a courseware generation method, an electronic device and a computer storage medium, to generate courseware with strong professional, accurate and interactive characteristics, and accurately meet the actual teaching needs. The above technical solution is as follows:

[0006] In a first aspect, the embodiments of the present application provide a courseware generation method, comprising:

[0007] receiving target creation information input by a user on a courseware creation page; the target creation information includes creation requirement information and / or a creation reference document;

[0008] obtaining an initial creation intention corresponding to the target creation information based on a large language model and a private domain knowledge base; the private domain knowledge base is a knowledge base constructed by preprocessing knowledge texts of a preset field;

[0009] displaying an intention editing page of the initial creation intention;

[0010] After obtaining a target creation intention edited by a user based on the initial creation intention, obtaining an initial creation outline corresponding to the target creation intention based on a large language model, a private domain knowledge base and an interactive tool library; the interactive tool library includes a plurality of interactive capability tools for providing interactive information;

[0011] displaying an outline editing page of the initial creation outline;

[0012] After obtaining the target creation outline edited by the user based on the initial creation outline, the courseware corresponding to the target creation outline is generated based on the large language model, the private domain knowledge base, the interactive tool library and the multi-modal material library.

[0013] The courseware display page displays the courseware, and the courseware includes at least one interactive ability tool provided by the interactive tool library.

[0014] In a possible implementation, the initial creation intention corresponding to the target creation information is obtained based on the large language model and the private domain knowledge base, including: inputting the target creation information into the large language model to obtain input understanding information; processing the input understanding information based on the private domain knowledge base to obtain context information corresponding to the input understanding information; and sending the input understanding information and the context information to the large language model to obtain the initial creation intention corresponding to the target creation information.

[0015] In a possible implementation, in a case where the target creation information includes creation requirement information and a creation reference document, the input understanding information is obtained by inputting the target creation information into the large language model, including: analyzing the creation reference document, and inputting reference document information obtained by the analysis and the creation requirement information into the large language model to obtain the input understanding information.

[0016] In a possible implementation, the initial creation outline corresponding to the target creation intention is obtained based on the large language model, the private domain knowledge base and the interactive tool library, including: inputting the target creation intention, the context information and tool description information of each interactive tool in the interactive tool library into the large language model to obtain the initial creation outline corresponding to the target creation intention.

[0017] In a possible implementation, the courseware corresponding to the target creation outline is generated based on the large language model, the private domain knowledge base, the interactive tool library and the multi-modal material library, including: creating at least one instruction generation task corresponding to the target creation outline according to the target creation outline; different instruction generation tasks correspond to different outline nodes in the target creation outline; obtaining courseware generation instructions corresponding to each instruction generation task based on the large language model, the context information, the interactive tool library and the multi-modal material library; respectively performing instruction calibration on the courseware generation instructions corresponding to each instruction generation task; calling the interactive tool library and the multi-modal material library, and generating the courseware corresponding to the target creation outline according to the calibrated courseware generation instructions corresponding to each instruction generation task.

[0018] In a possible implementation, the courseware generation instruction corresponding to each instruction generation task is obtained based on the large language model, the context information, the interactive tool library, and the multi-modal material library, including: inputting the following information into the large language model: the outline text of the outline node corresponding to each instruction generation task, the context information, the tool description information of each interactive tool in the interactive tool library, the set of material tags corresponding to each courseware generation material in the multi-modal material library, and the set of preset layout formats of the courseware; and obtaining the courseware generation instruction corresponding to each instruction generation task output by the large language model.

[0019] In a possible implementation, the private domain knowledge base is constructed in the following manner: dividing the knowledge text of a preset domain according to a preset division rule to obtain at least two divided segments; performing text vectorization processing on the at least two divided segments to obtain text vectorization data corresponding to the at least two divided segments; constructing a vector library according to the text vectorization data corresponding to the at least two divided segments; the vector library is used for analyzing input understanding information; constructing a business structural database according to the knowledge text of the preset domain; the business structural database has a mapping relationship with the content in the vector library; the business structural database is used for searching the input understanding information; and constructing the private domain knowledge base based on the vector library and the business structural database.

[0020] In a possible implementation, the multi-modal material library is constructed in the following manner: obtaining courseware generation materials of multiple different modalities; the courseware generation materials of multiple different modalities include one or more of picture materials, video materials, and interactive materials; performing tag processing on the courseware generation materials of multiple different modalities by using a multi-modal large model to obtain material tags corresponding to the courseware generation materials of multiple different modalities; and constructing the multi-modal material library according to the courseware generation materials of multiple different modalities and the material tags corresponding to the courseware generation materials of multiple different modalities.

[0021] In a possible implementation, the input understanding information is processed based on the private domain knowledge base to obtain context information corresponding to the input understanding information, including: searching the input understanding information in the private domain knowledge base to obtain first search information; searching the input understanding information through the Internet to obtain second search information; and combining the first search information and the second search information to obtain the context information corresponding to the input understanding information.

[0022] In a possible implementation, the intention editing page further displays an intention editing control, which is configured to receive an editing operation of the user on the initial creation intention.

[0023] In a possible implementation, the initial creation outline includes a plurality of end nodes, and each end node corresponds to a node editing control, which is configured to receive an editing operation of the user on the outline text included in the end node.

[0024] In a possible implementation, the outline editing page further displays at least two outline display form selection controls respectively used for controlling the initial creation outline to be displayed in different display forms.

[0025] In a possible implementation, the courseware display page includes a display area of the target creation outline and a preview area of the courseware; the target creation outline includes a plurality of outline nodes, and the outline text included in each of the outline nodes corresponds to the courseware content of the courseware; the courseware generation method further includes: receiving a fourth trigger operation of a user on a target outline node; the target outline node is any one of the plurality of outline nodes; and in response to the fourth trigger operation, displaying the courseware content corresponding to the outline text included in the target outline node in the preview area of the courseware.

[0026] In a possible implementation, the display area of the target creation outline includes the target creation outline and at least two outline display form selection controls, and the at least two outline display form selection controls are respectively used for controlling the target creation outline to be displayed in different display forms.

[0027] In a second aspect, the embodiments of the present application provide a courseware generation method, including:

[0028] inputting the target creation information into the large language model to obtain input understanding information;

[0029] processing the input understanding information based on a private domain knowledge base to obtain context information corresponding to the input understanding information; the private domain knowledge base is a knowledge base constructed by preprocessing knowledge texts of a preset domain, and the target creation information includes creation requirement information and / or a creation reference document;

[0030] sending the input understanding information and the context information to the large language model to obtain an initial creation intent corresponding to the target creation information;

[0031] after receiving the target creation intent, inputting the target creation intent, the context information, and tool description information of each interactive tool in an interactive tool library into the large language model to obtain an initial creation outline corresponding to the target creation intent; the target creation intent is edited by a user based on the initial creation intent, and the interactive tool library includes a plurality of interactive capability tools for providing interactive information;

[0032] after receiving the target creation outline, generating a courseware corresponding to the target creation outline based on the target creation outline, the private domain knowledge base, the interactive tool library, and a multi-modal material library; the target creation outline is edited by a user based on the initial creation outline.

[0033] In a possible implementation, based on the target creation outline, the private domain knowledge base, the interactive tool library and the multi-modal material library, a courseware corresponding to the target creation outline is generated, including: creating at least one instruction generation task corresponding to the target creation outline according to the target creation outline; different instruction generation tasks correspond to different outline nodes in the target creation outline; based on the large language model, the context information, the interactive tool library and the multi-modal material library, obtaining courseware generation instructions corresponding to each instruction generation task; respectively calibrating the courseware generation instructions corresponding to each instruction generation task; calling the interactive tool library and the multi-modal material library, and generating the courseware corresponding to the target creation outline according to the calibrated courseware generation instructions corresponding to each instruction generation task.

[0034] In a possible implementation, based on the large language model, the context information, the interactive tool library and the multi-modal material library, the courseware generation instructions corresponding to each instruction generation task are obtained, including: inputting the following information into the large language model: the outline text of the outline node corresponding to each instruction generation task, the context information, the tool description information of each interactive tool in the interactive tool library, the set of material tags corresponding to each courseware generation material in the multi-modal material library, and the set of preset layout formats of the courseware; obtaining the courseware generation instructions corresponding to each instruction generation task output by the large language model.

[0035] In a possible implementation, the target creation information is input into the large language model to obtain input understanding information, including: analyzing the creation reference document, inputting the analyzed reference document information and the creation requirement information into the large language model to obtain the input understanding information.

[0036] In a possible implementation, the private domain knowledge base is constructed in the following manner: dividing the knowledge text of a preset field according to a preset division rule to obtain at least two divided segments; performing text vectorization processing on the at least two divided segments to obtain text vectorization data corresponding to the at least two divided segments; constructing a vector library according to the text vectorization data corresponding to the at least two divided segments; the vector library is used for analyzing input understanding information; constructing a business structural database according to the knowledge text of the preset field; the business structural database has a mapping relationship with the content in the vector library; the business structural database is used for searching the input understanding information; and the private domain knowledge base is constructed based on the vector library and the business structural database.

[0037] In a possible implementation, the multi-modal material library is constructed by: obtaining courseware generation materials of multiple different modalities; the courseware generation materials of multiple different modalities include one or more of picture materials, video materials, and interactive materials; performing labelization processing on the courseware generation materials of multiple different modalities by using a multi-modal large model to obtain material labels corresponding to the courseware generation materials of multiple different modalities; and constructing the multi-modal material library according to the courseware generation materials of multiple different modalities and the material labels corresponding to the courseware generation materials of multiple different modalities.

[0038] In a third aspect, the embodiments of the present application provide a courseware generation method, including:

[0039] displaying a courseware creation page;

[0040] receiving target creation information input by a user on the courseware creation page; the target creation information includes creation requirement information and / or a creation reference document;

[0041] displaying an intention editing page, the intention editing page displaying an initial creation intention corresponding to the target creation information; the initial creation intention is obtained based on a large language model and a private domain knowledge base;

[0042] receiving a first editing operation of the user on the initial creation intention;

[0043] displaying an outline editing page, the outline editing page displaying an initial creation outline corresponding to the target creation intention, the target creation intention being obtained after the initial creation intention is edited according to the first editing operation, and the initial creation outline being obtained based on the large language model, the private domain knowledge base, and an interactive tool library, and the interactive tool library including multiple interactive capability tools for providing interactive information;

[0044] receiving a second editing operation of the user on the initial creation outline;

[0045] displaying a courseware display page, the courseware display interface displaying a courseware corresponding to the target creation outline, the target creation outline being obtained after the initial creation outline is edited according to the second editing operation, and the courseware being obtained based on the large language model, the private domain knowledge base, the interactive tool library, and a multi-modal material library, and the courseware including at least one interactive capability tool provided by the interactive tool library.

[0046] In a possible implementation, the initial creation intention is obtained by:

[0047] inputting the target creation information into the large language model to obtain input understanding information;

[0048] processing the input understanding information based on the private domain knowledge base to obtain context information corresponding to the input understanding information;

[0049] The input understanding information and the context information are sent to the large language model to obtain an initial creation intention corresponding to the target creation information.

[0050] In a possible implementation, in a case where the target creation information includes creation requirement information and a creation reference document, the target creation information is input to the large language model to obtain input understanding information, including:

[0051] The creation reference document is parsed, and the parsed reference document information and the creation requirement information are input to the large language model to obtain the input understanding information.

[0052] In a possible implementation, based on the large language model, the private domain knowledge base, and the interactive tool library, an initial creation outline corresponding to the target creation intention is obtained, including:

[0053] The target creation intention, the context information, and tool description information of each interactive tool in the interactive tool library are input to the large language model to obtain the initial creation outline corresponding to the target creation intention.

[0054] In a possible implementation, the courseware is obtained in the following manner:

[0055] According to the target creation outline, at least one instruction generation task corresponding to the target creation outline is created; different instruction generation tasks correspond to different outline nodes in the target creation outline.

[0056] Based on the large language model, the context information, the interactive tool library, and the multi-modal material library, courseware generation instructions corresponding to each instruction generation task are obtained.

[0057] The courseware generation instructions corresponding to each instruction generation task are respectively calibrated.

[0058] The interactive tool library and the multi-modal material library are called, and the courseware corresponding to the target creation outline is generated according to the calibrated courseware generation instructions corresponding to each instruction generation task.

[0059] In a possible implementation, based on the large language model, the context information, the interactive tool library, and the multi-modal material library, the courseware generation instructions corresponding to each instruction generation task are obtained, including:

[0060] The following information is input to the large language model: the outline text of the outline node corresponding to each instruction generation task, the context information, the tool description information of each interactive tool in the interactive tool library, the set of material tags corresponding to each courseware generation material in the multi-modal material library, and the set of preset layout formats of the courseware.

[0061] The courseware generation instructions corresponding to each instruction generation task output by the large language model are obtained.

[0062] In a possible implementation, the private domain knowledge base is constructed in the following manner:

[0063] The knowledge text of the preset domain is divided according to a preset division rule, to obtain at least two division segments;

[0064] The at least two division segments are subjected to text vectorization processing, to obtain text vectorization data corresponding to the at least two division segments;

[0065] A vector library is constructed according to the text vectorization data corresponding to the at least two division segments; the vector library is used for analyzing input understanding information;

[0066] A business structural database is constructed according to the knowledge text of the preset domain; the business structural database has a mapping relationship with the content in the vector library; the business structural database is used for searching the input understanding information;

[0067] The private domain knowledge base is constructed based on the vector library and the business structural database.

[0068] In a possible implementation, the multi-modal material library is constructed in the following manner:

[0069] Obtain courseware generation materials of multiple different modalities; the courseware generation materials of multiple different modalities include one or more of picture materials, video materials, and interactive materials;

[0070] The courseware generation materials of multiple different modalities are subjected to labelization processing by using a multi-modal large model, to obtain material labels corresponding to the courseware generation materials of multiple different modalities;

[0071] The multi-modal material library is constructed according to the courseware generation materials of multiple different modalities and the material labels corresponding to the courseware generation materials of multiple different modalities.

[0072] In a possible implementation, the input understanding information is processed based on the private domain knowledge base, to obtain context information corresponding to the input understanding information, including:

[0073] The input understanding information is searched in the private domain knowledge base, to obtain first search information;

[0074] The input understanding information is searched through the Internet, to obtain second search information;

[0075] The first search information and the second search information are combined to obtain the context information corresponding to the input understanding information.

[0076] In a possible implementation, the initial creation outline includes multiple end nodes, each end node is followed by a node editing control, and the node editing control is used for receiving an editing operation of a user on outline text included in the end node.

[0077] In a possible implementation, the outline editing page further displays at least two outline display form selection controls respectively used for controlling the initial creation outline to be displayed in different display forms.

[0078] In a possible implementation, the courseware display page includes a display area of the target creation outline and a preview area of the courseware.

[0079] The target creation outline includes a plurality of outline nodes, and the outline text included in each of the outline nodes corresponds to the courseware content of the courseware.

[0080] The fourth trigger operation of the user is received for the target outline node, and the target outline node is any one of the plurality of outline nodes.

[0081] In response to the fourth trigger operation, the courseware content corresponding to the outline text included in the target outline node is displayed in the preview area of the courseware.

[0082] In a possible implementation, the display area of the target creation outline includes the target creation outline and at least two outline display form selection controls respectively used for controlling the target creation outline to be displayed in different display forms.

[0083] In a fourth aspect, an electronic device is provided, including a processor and a memory; the memory stores a computer program, and the computer program is adapted to be loaded by the processor and execute the method steps provided in the first aspect or the second aspect.

[0084] In a fifth aspect, a computer storage medium is provided, which stores a plurality of instructions, and the instructions are adapted to be loaded by a processor and execute the method steps provided in the first aspect or the second aspect.

[0085] The courseware generation method, the electronic device and the computer storage medium can effectively solve the problems that the courseware generated in the prior art lacks pertinence and accuracy, the courseware content is random, and the actual teaching needs cannot be accurately met by receiving target creation information input by a user on a courseware creation page, combining a large language model, a private domain knowledge base, an interactive tool library and a multi-modal material library to generate courseware. Specifically, by introducing the private domain knowledge base constructed based on the knowledge text of the preset field, it can not only ensure that the content of the generated courseware conforms to the language logic, but also assist in verifying the content of the courseware according to the private domain professional knowledge, so as to accurately meet the specific requirements of a specific educational institution, a subject or a teaching stage, thereby ensuring the pertinence and accuracy of the courseware. By introducing the interactive tool library including a plurality of interactive ability tools and the multi-modal material library including a plurality of courseware generation materials of different modalities to participate in the courseware design, the interactivity and richness of the courseware content are improved, thereby ensuring that the generated courseware can effectively stimulate the active thinking and in-depth exploration of students and accurately meet the actual teaching needs. In addition, the generation of the courseware does not simply rely on the direct generation capability of the large language model end to end, but divides the generation process of the courseware into three core stages of creation intention generation, creation outline generation and final courseware generation. In the process of generating the courseware, the user can also receive editing of the initial creation intention and the initial creation outline, so that the user can participate in and control the entire courseware creation process, reducing the randomness of the courseware content and enhancing the controllability of the courseware content. The courseware generated by the scheme can ensure that the finally generated courseware has pertinence, accuracy and controllability, and accurately meets the actual teaching needs. BRIEF DESCRIPTION OF DRAWINGS

[0086] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0087] FIG. 1 is a flow diagram of a courseware generation method according to an example embodiment of the present application;

[0088] FIG. 2 is a schematic diagram of a courseware creation page according to an example embodiment of the present application;

[0089] FIG. 3 is a schematic diagram of an intention editing page according to an example embodiment of the present application;

[0090] FIG. 4 is a schematic diagram of an outline editing page according to an example embodiment of the present application;

[0091] FIG. 5 is a schematic diagram of a courseware display page according to an example embodiment of the present application;

[0092] FIG. 6 is a flow diagram of another courseware generation method according to an example embodiment of the present application;

[0093] FIG. 7 is an architecture diagram of obtaining input understanding information according to an example embodiment of the present application;

[0094] FIG. 8 is a flow diagram of a private domain knowledge base construction method according to an example embodiment of the present application;

[0095] FIG. 9 is a flow diagram of a multi-modal material library construction method according to an example embodiment of the present application;

[0096] FIG. 10 is a structural diagram of a courseware generation device according to an example embodiment of the present application;

[0097] FIG. 11 is a structural diagram of another courseware generation device according to an example embodiment of the present application;

[0098] FIG. 12 is a structural diagram of an electronic device according to an example embodiment of the present application;

[0099] FIG. 13 is a structural diagram of another electronic device according to an example embodiment of the present application. DETAILED DESCRIPTION

[0100] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and should not be used to limit the present application.

[0101] In the description of the present application, it should be understood that the terms "first", "second" and the like are only for the purpose of description and should not be understood as indicating or implying relative importance. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances. In addition, in the description of the present application, "multiple" means two or more, unless otherwise specified. "And / or", which describes the relationship between the associated objects, means that there can be three relationships, for example, A and / or B can mean that there are three cases of A alone, A and B together, and B alone. The character " / " generally represents that the associated objects before and after are in an "or" relationship.

[0102] At present, the generation of courseware by using artificial intelligence technology often only relies on large language models. Large language models (LLM for short) refer to deep learning models trained using a large amount of text data, which can generate natural language text or understand the meaning of language text. Large language models can handle a variety of natural language tasks such as text classification, question answering, and dialogue, and are an important way to artificial intelligence. In existing solutions, large language models on the Internet can be directly applied to complete some functions of artificial intelligence without the need to locally build and train complex models. In daily chat, simple dialogue, and text summary natural language to large language models, only relying on large language models to generate content can have better results. However, if the courseware is completely generated based on large language models, only the content of the generated courseware can conform to the language logic, and the quality of the generated courseware cannot be guaranteed. Specifically, the biggest problem of generating courseware by using the existing technical solution is that only by inputting simple information into the large language model, the large language model searches and generates courseware on the network, only the language logic of the courseware can be guaranteed to be correct, and there may be errors or inaccurate information in the generated courseware content, resulting in low accuracy of the output courseware, poor pertinence, and the need for users to make more adjustments, poor usability. In order to solve the above problems, one improvement direction of the existing solution is to increase the interaction between the data input into the large language model and the large language model, and the large language model obtains the corresponding courseware outline according to the user input information, and the user adjusts the outline, and then generates the corresponding courseware. However, due to the uncertainty of the large language model, the generated outline may also need to be adjusted, and the usability is not strong. Another improvement direction of the existing solution is to construct a local large language model, and train the large language model locally using relevant materials in the education field, so that the finally trained large language model can generate courseware with good pertinence and high accuracy. However, this way requires a powerful algorithm team, and a lot of manpower and material resources are needed to complete the training of the large model, which is too difficult for ordinary users and has low usability.

[0103] To solve the above problems, the application provides a new solution, a courseware generation method, an electronic device and a computer storage medium. After receiving the target creation information input by the user on the courseware creation page, a private domain knowledge base constructed based on a preset domain knowledge text is introduced, which can not only ensure that the generated courseware content conforms to the language logic, but also can assist in checking the courseware content according to the private domain professional knowledge, so as to accurately meet the specific requirements of specific educational institutions, disciplines or teaching stages, and ensure the pertinence and accuracy of the courseware. By introducing an interactive tool library including a plurality of interactive ability tools, and a multi-modal material library including a plurality of courseware generation materials of different modalities to participate in courseware design, the interactivity and richness of the courseware content are improved, so that the generated courseware can effectively stimulate the active thinking and in-depth exploration of students, and accurately meet the actual teaching needs. In addition, the generation of the above courseware does not simply rely on the direct generation ability of the large language model end-to-end, but divides the generation process of the courseware into three core stages of creation intention generation, creation outline generation and final courseware generation. In the process of generating the courseware, the user can also receive editing of the initial creation intention and the initial creation outline, so that the user can participate in and control the entire courseware creation process, reducing the randomness of the courseware content and enhancing the controllability of the courseware content.

[0104] The above scheme provided by the application can effectively solve the problems in the prior art that the generated courseware lacks pertinence and accuracy, the courseware content is random, and cannot accurately meet the actual teaching needs, and can ensure that the finally generated courseware has pertinence, accuracy and controllability, and accurately meets the actual teaching needs. Moreover, the above scheme only needs to construct a private domain knowledge base, an interactive tool library and a multi-modal material library, so that the existing large language model can be directly used for courseware generation, and the editing operation of the user on the creation intention and the creation outline during the courseware generation process can be received, so that a courseware with strong pertinence, high accuracy and strong usability can be obtained at low investment cost.

[0105] In one embodiment, as shown in FIG. 1, a courseware generation method is provided. In this embodiment, the method is applied to a terminal. It should be understood that the method can also be applied to a system including a terminal and a server, and can be realized through the interaction of the terminal and the server. In this embodiment, the method includes the following steps:

[0106] S101: Display a courseware creation page, and the courseware creation page displays a first trigger control.

[0107] The courseware creation page is an interactive page for a user to input target creation information. The first trigger control includes, but is not limited to, an information sending button in the courseware creation page, and is used to send the target creation information input by the user to the terminal to generate an initial creation intention in response to a first trigger operation of the user.

[0108] For example, referring to FIG. 2, the courseware creation page 20 displays an information input area 21 and an input recommendation area 22. The information input area 21 is used to receive the target creation information input by the user, and the first trigger control is an information sending button 23 in the information input area 21.

[0109] S102: Receive the target creation information input by the user in the courseware creation page.

[0110] The target creation information includes creation requirement information and / or creation reference documents. The creation requirement information includes, but is not limited to, course type selection information and text prompt information input by the user in the courseware creation page, and the creation reference documents include, but are not limited to, text documents, picture documents, audio and video documents, and rich media documents.

[0111] Optionally, the user can input course type selection information 24, input text prompt information 25, and upload creation reference documents 26 in the information input area 21 of the courseware creation page 20 as shown in FIG. 2. In addition, the user can also refer to the input recommendation items in the input recommendation area 22 to input the text prompt information 25, so that the target creation information input is clearer and more specific.

[0112] For example, the terminal receives the course type selection information 24 of “new teaching” and the input text prompt information 25 of “please create a courseware about the course text of ‘Jingyang Ridge’ according to the uploaded document and in combination with your ability” input by the user in the information input area 21 of the courseware creation page 20 as shown in FIG. 2, and the Word document of Jingyang Ridge.

[0113] S103: In response to the first trigger operation on the first trigger control, an initial creation intention corresponding to the target creation information is obtained based on a large language model and a private domain knowledge base. The private domain knowledge base is a knowledge base constructed by preprocessing knowledge texts in a preset field.

[0114] The first trigger operation includes but is not limited to a custom interaction action such as clicking, sliding, long pressing, etc. The private domain knowledge base refers to a preset domain knowledge base constructed for an individual, an institution or an organization, and the target creation information belongs to the preset domain. For example, the private domain knowledge base can be an education domain knowledge base constructed for school teachers, which covers various subject curriculum standards, teacher's books, student's books and other teaching materials, and the target creation information can be course type selection information, text prompt information and various types of reference documents input by school teachers for preparing teaching courseware.

[0115] Optionally, after the user inputs the target creation information in the courseware creation page 20 as shown in FIG. 2 and clicks the information sending button 23, the terminal responds to the user's clicking operation and acquires the initial creation intent corresponding to the target creation information based on the locally integrated large language model and the private domain knowledge base constructed for school teachers in the education domain. In this embodiment, by combining the private domain knowledge base containing rich teaching materials, the limitation of the general large language model in the specific domain knowledge is overcome, ensuring that the generated content not only conforms to the language logic, but also meets the specific requirements of the specific education institution, subject or teaching stage, improving the professionalism and accuracy of the model generated content.

[0116] Optionally, the step of acquiring the initial creation intent corresponding to the target creation information based on the large language model and the private domain knowledge base can also not be triggered to execute according to the first trigger operation of the first trigger control, but automatically executed after a period of time after automatically acquiring the target creation information input by the user.

[0117] S104: Display an intent editing page of the initial creation intent, and the intent editing page displays a second trigger control; the initial creation intent is editable.

[0118] The intent editing page is an interactive page for the user to edit and confirm the initial creation intent. The second trigger control includes but is not limited to a generate outline button in the intent editing page, which is used to send the target creation intent confirmed by the user to the terminal in response to the second trigger operation of the user, so as to generate an initial creation outline corresponding to the target creation intent.

[0119] Optionally, the intent editing page also displays an intent editing control for receiving the editing operation of the user on the initial creation intent. The intent editing control includes but is not limited to an intent adding control, an intent deleting control, an intent modifying control, etc.; the intent editing control can be an independent button on the intent editing page, or can be integrated in a certain area in the intent editing page.

[0120] Exemplarily, referring to FIG. 3, the intention editing page 30 includes three intention display areas 31-33 for displaying the initial creative intentions, a generation outline button 34, and an intention adding button 35. Based on the intention editing page 30, the user can perform editing operations such as adding, deleting, modifying, and the like on the initial creative intentions. For example, the user can click the intention adding button 35 below the intention display area 33 to add a creative intention; can perform a leftward sliding operation based on the intention display area 33 to delete the initial creative intention displayed in the intention display area 33; and can click the intention display area 33 to enter a modification mode to modify the initial creative intention displayed in the intention display area 33.

[0121] It should be noted that the embodiment only describes one preferred embodiment of the present application, and the initial creative intentions can be displayed in different areas of the intention editing page in different columns or can be displayed in the same area of the intention editing page. The user can click the intention adding control to add a creative intention or can add a creative intention by sliding, long-pressing, or clicking other areas of the intention editing page. The specific display manner of the initial creative intentions and the specific layout manner of the intention editing controls are not limited in the embodiment of the present application.

[0122] S105: In response to a second triggering operation on the second triggering control, after obtaining the target creative intention edited by the user based on the initial creative intention, an initial creative outline corresponding to the target creative intention is obtained based on the large language model, the private domain knowledge base, and the interactive tool library. The interactive tool library includes a plurality of interactive ability tools for providing interactive information. Here, the interactive information can be a simple information display interface or an interactive interface with a man-machine interaction interface, such as a game interface, a question answering interface, and the like.

[0123] The second triggering operation includes but is not limited to a self-defined interactive action such as clicking, sliding, long-pressing, and the like. The interactive tool library includes a plurality of interactive ability tools for human-computer interaction, such as classroom activities, classical poems, mind maps, word cards, and the like. Through the interactive ability tools, the finally generated courseware can also realize human-computer interaction with teachers or students, and improve the diversified results of the automatically generated courseware.

[0124] The interactive ability tool of the classroom activity can include an answering activity tool, which is embodied as a selection question, a judgment question, and the like. The interactive ability tool of the word card can perform activities such as word spelling. Optionally, the interactive ability tool is described by tool description information, which specifically can include a tool tag and a function definition. The interactive ability tool is identified by the tool tag, and the tool capability of the specific interactive ability tool is defined by the function definition. For example, the interactive ability tool of the ancient poetry can be identified by the tags "ancient poetry" and "poetry", and the specific function is defined and described by the programming function, so that when the interactive ability tool is called, the corresponding function can be implemented. In an embodiment, the collected interactive ability tools with existing function definitions can be tagged, so as to form the function description and the corresponding tags of each interactive ability tool into the tool description information of each interactive ability tool, and build an interactive tool library.

[0125] Optionally, after the user completes the editing of the initial creation intention in the intention editing page 30 as shown in FIG. 3, clicks the generate outline button 34, and the terminal responds to the click operation of the user, the target creation intention corresponding to the initial creation outline is obtained based on the locally integrated large language model, the private domain knowledge base, and the interactive tool library including the classroom activity, the ancient poetry, the mind map, the word card, and the like, so that the generated initial creation outline has the interactive ability tool identifier. In this embodiment, by introducing the interactive tool library containing various interactive ability tools to participate in the outline generation and courseware design, the interactivity of the courseware is enhanced, the richness of the courseware content is improved, and the teaching effect is further improved.

[0126] S106: display an outline editing page of the initial creation outline, and the outline editing page displays a third trigger control; the initial creation outline includes a plurality of outline nodes, each outline node includes an outline text, and at least one outline node has an identifier of an interactive ability tool corresponding to the outline node. Specifically, the identifier of the interactive ability tool can be displayed after the outline text of the outline node, or can be displayed above, below, or the like of the outline text, or can be displayed after the input focus is slid to the outline text.

[0127] The outline editing page is an interactive page for the user to edit and confirm the initial creation outline. The third trigger control includes, but is not limited to, a courseware generation button in the outline editing page, and is used to send the target creation outline confirmed by the user to the terminal in response to the third trigger operation of the user, so as to generate the courseware corresponding to the target creation outline.

[0128] Exemplarily, please refer to FIG. 4, the initial creation outline 41 is displayed in the form of a guide map in the outline editing page 40. The initial creation outline 41 includes a plurality of outline nodes, each of which includes an outline text corresponding to the courseware content to be generated, i.e., the plurality of outline nodes of the initial creation outline correspond to the courseware content to be generated. After the three outline texts of "learning flashcard", "class activity selection question link", and "mind map teaching", the identifiers 42-44 of the interactive ability tools are displayed respectively, for prompting that the courseware content to be generated corresponding to the above-mentioned outline nodes will display the three interactive ability tools of learning flashcard, class activity, and mind map.

[0129] Optionally, the outline editing page further displays at least two outline display form selection controls for controlling the initial creation outline to be displayed in different display forms. The outline display form selection controls include, but are not limited to, a guide map form selection control, a list form selection control, etc.; the outline display form selection controls can be independent buttons on the outline editing page, or can be integrated in a certain region in the intent editing page. Please refer to FIG. 4, the guide map form selection button 45 and the list form selection button 46 are further displayed in the outline editing page 40. The user can switch the display form of the initial creation outline by clicking the guide map form selection button 45 or the list form selection button 46. It is worth noting that the present embodiment only describes one preferred embodiment of the present application, the user can click the outline display form selection controls such as the guide map form selection button 45, the list form selection button 46, or slide, long press, or click other regions on the outline editing page to switch the display form of the initial creation outline. The specific layout of the outline display form selection controls is not limited in the present embodiment.

[0130] In the present embodiment, by providing a plurality of outline display form selection controls, the user can select the display form of the initial creation outline according to personal preferences and actual needs, so as to facilitate the user to more intuitively understand and edit the initial creation outline, and improve the efficiency of courseware generation.

[0131] Optionally, the initial creation outline includes a plurality of end nodes, and each end node corresponds to a node editing control behind it, which is used to receive the editing operation of the user on the outline text included in the end node.

[0132] Among them, the end node refers to the lowest level of the outline node, which has no further sub-level content. The node editing control includes, but is not limited to, an outline node adding control, an outline node deleting control, an outline node modifying control, etc.; the node editing control can be an independent button on the outline editing page, or can be integrated in a certain region in the outline editing page.

[0133] It can be understood that the user can not only edit the outline text included in the end node, but also add, delete, modify and edit the initial creation outline 41. For example, the user can click the outline node addition control after the outline node to add a child outline node to the outline node; long press the outline node to delete the outline node; and click the outline text in the outline node to enter the modification mode to modify the outline text in the outline node.

[0134] In this embodiment, by setting the node editing control after each end node of the initial creation outline, the user can directly edit the initial creation outline generated by the large language model according to the actual teaching needs, so as to quickly generate a courseware that meets the user's requirements. Not only does it enhance the controllability of the courseware generation process, but also improves the efficiency of courseware generation.

[0135] S107: In response to a third trigger operation on the third trigger control, the target creation outline edited by the user based on the initial creation outline is obtained, and a courseware corresponding to the target creation outline is generated based on the large language model, the private knowledge base, the interactive tool library and the multi-modal material library. The multi-modal material library includes a plurality of different modal courseware generation materials.

[0136] The third trigger operation includes but is not limited to a custom interaction action such as clicking, sliding, long pressing, etc. The interactive tool library includes a plurality of interactive ability tools such as classroom activities, classical poems, mind maps, word cards, etc.

[0137] Optionally, this step can also not need to be triggered and executed according to the third trigger operation of the third trigger control, but can be automatically executed after a period of time after receiving the user's editing operation on the initial creation outline, or can be automatically executed after waiting for a period of time to determine that the user's editing operation on the initial creation outline is not received.

[0138] Optionally, after the user completes the editing of the initial creation outline in the outline editing page 40 as shown in FIG. 4 and clicks the courseware generation button 47, the terminal generates a courseware corresponding to the target creation outline based on the locally integrated large language model, private knowledge base, interactive tool library and multi-modal material library in response to the user's click operation.

[0139] In this embodiment, on the basis of a large language model, a private domain knowledge base is constructed by introducing a preset domain-based knowledge text for preprocessing to generate courseware, ensuring that the generated content not only conforms to language logic, but also meets the specific requirements of a specific educational institution, discipline, or teaching stage, improving the professionalism and accuracy of the courseware. By introducing an interactive tool library including a variety of interactive tools and a multi-modal material library including a variety of different modal courseware generation materials to participate in courseware design, the interactivity and richness of the courseware content are improved, thereby ensuring that the generated courseware can effectively stimulate students' active thinking and in-depth exploration, accurately meet the actual teaching needs, and thus improve the teaching effect.

[0140] S108: Display a courseware display page including the courseware, the courseware including at least one interactive ability tool provided by the interactive tool library.

[0141] The courseware display page is an interactive page for the user to quickly browse and confirm the generated courseware.

[0142] Optionally, the courseware display page includes a display area of the target creation outline and a preview area of the courseware; the target creation outline includes a plurality of outline nodes, and the outline text included in each outline node corresponds to the courseware content of the courseware; the courseware generation method further includes: receiving a fourth trigger operation of the user on the target outline node; the target outline node is any one of the plurality of outline nodes; in response to the fourth trigger operation, displaying the courseware content corresponding to the outline text included in the target outline node in the preview area of the courseware.

[0143] The fourth trigger control includes but is not limited to a self-defined interactive action such as clicking, sliding, and long pressing on each outline node in the target creation outline.

[0144] For example, referring to FIG. 5, the courseware display page 50 includes a display area 51 of the target creation outline on the left side of the page and a preview area 52 of the courseware on the right side of the page. The target creation outline includes a plurality of outline nodes, and the outline text included in each outline node corresponds to the courseware content of the courseware. If the user clicks on the outline node 53 corresponding to the outline text "recognize key words in the text" in the display area 51 of the target creation outline, the preview area 52 of the courseware displays the "recognize key words in the text" page 54 corresponding to the outline node.

[0145] In this embodiment, the target creation outline and the finally generated courseware are displayed on the courseware display page, so that the user can view the courseware content corresponding to each outline node by a simple triggering operation, preview the generated courseware quickly, and quickly locate the content of each part in the courseware according to the target creation outline, so as to evaluate the effect of the courseware content corresponding to each outline node in the target creation outline, and provide an efficient and convenient courseware preview mode for the user.

[0146] Optionally, the display area of the target creation outline comprises the target creation outline and at least two outline display form selection controls, and the at least two outline display form selection controls are respectively used to control the target creation outline to be displayed in different display forms.

[0147] Please refer to FIG. 5. The display area 51 of the target creation outline in the courseware display page 50 displays a guide map form selection button 55 and a list form selection button 56. The user can switch the display form of the initial creation outline by clicking the guide map form selection button 55 or the list form selection button 56.

[0148] In this embodiment, by providing a plurality of outline display form selection controls, the user can select the display form of the target creation outline according to personal preferences and actual needs, so as to facilitate the user to preview and understand the target creation outline more intuitively, enhance the interactivity in the courseware generation process, and improve the user experience.

[0149] The courseware generation method can effectively solve the problems that the courseware generated in the prior art lacks pertinence and accuracy, the courseware content is random, and the actual teaching needs cannot be accurately met. Specifically, by introducing the private domain knowledge base constructed based on the preset field knowledge text, it can not only ensure that the generated courseware content conforms to the language logic, but also assist in verifying the courseware content according to the private domain professional knowledge, so as to accurately meet the specific requirements of a specific educational institution, a subject or a teaching stage, and ensure the pertinence and accuracy of the courseware. By introducing the interactive tool library including a plurality of interactive tools and the multi-modal material library including a plurality of courseware generation materials of different modalities to participate in courseware design, the interactivity and richness of the courseware content are improved, so that the generated courseware can effectively stimulate the active thinking and in-depth exploration of students, and accurately meet the actual teaching needs. In addition, the generation of the courseware is not simply dependent on the direct generation capability of the large language model, but is divided into three core stages of creation intention generation, creation outline generation and final courseware generation. In the process of generating the courseware, by displaying the intention editing page of the initial creation intention and the outline editing page including the initial creation outline, the user can participate in and control the entire courseware creation process, reducing the randomness of the courseware content and enhancing the controllability of the courseware content. By using the above method to generate courseware, it can ensure that the finally generated courseware has pertinence, accuracy and controllability, and accurately meets the actual teaching needs.

[0150] In one embodiment, as shown in FIG. 6, another courseware generation method is provided. In this embodiment, the method is applied to the server as an example. It can be understood that the method can also be applied to the terminal, and can also be applied to a system including a terminal and a server, and can be realized through the interaction of the terminal and the server. In this embodiment, the method includes the following steps:

[0151] S601: input the target creation information into the large language model to obtain input understanding information.

[0152] Optionally, the target creation information includes creation requirement information and / or creation reference document. In the case where the target creation information includes the creation requirement information and the creation reference document, inputting the target creation information into the large language model to obtain the input understanding information includes: parsing the creation reference document, inputting the parsed reference document information and the creation requirement information into the large language model to obtain the input understanding information.

[0153] The creation requirement information includes, but is not limited to, course type selection information and text prompt information input by the user in a courseware creation page, and the creation reference document includes, but is not limited to, a text document, a picture document, an audio / video document, and a rich media document (such as a PPT).

[0154] Referring to FIG. 7, in the case where the target creation information includes both the creation requirement information and the creation reference document, the server first parses the creation reference document uploaded by the user by using a content parsing engine to obtain document text information corresponding to the creation reference document, and then inputs the parsed document text information and the creation requirement information (including course type selection information and text prompt information) into a large language model to enable the large language model to understand and summarize the target creation information input by the user to obtain input understanding information. Understandably, if the target creation information only includes the creation reference document, the parsed document text information is directly input into the large language model to obtain the input understanding information after the creation reference document is parsed by using the content parsing engine; if the target creation information only includes the creation requirement information, the creation requirement information is directly input into the large language model to obtain the input understanding information.

[0155] In this embodiment, by jointly inputting the parsed reference document information and the creation requirement information into the large language model, it is ensured that the input understanding information meets the specific requirements of the user and is more consistent with the content of the reference document, so that the input understanding information is more professional.

[0156] S602: Process the input understanding information based on a private domain knowledge base to obtain context information corresponding to the input understanding information; the private domain knowledge base is a knowledge base constructed by preprocessing knowledge texts in a preset field.

[0157] The private domain knowledge base is a knowledge base constructed for an individual, an institution, or an organization in a preset field. It is constructed by knowledge division, text vectorization, and vector storage on knowledge texts in the preset field, and the target creation information belongs to the preset field. For example, the private domain knowledge base can be a knowledge base constructed for teachers in an education field, which covers various course standards, teacher's books, student's books, and other teaching materials in the education field, and the target creation information can be course type selection information, text prompt information, and various types of reference documents input by the teachers for preparing a courseware.

[0158] In this embodiment, the context information corresponding to the input understanding information is obtained based on the professional knowledge in the preset field provided by the private domain knowledge base, so that the content of the courseware generated subsequently not only meets the language logic, but also accurately meets the specific requirements of a specific educational institution, a subject, or a teaching stage, ensuring the pertinence and accuracy of the courseware.

[0159] Optionally, the input understanding information is processed based on the private domain knowledge base to obtain context information corresponding to the input understanding information, including: in the private domain knowledge base, the input understanding information is searched to obtain first search information; the input understanding information is searched through the Internet to obtain second search information; and the first search information and the second search information are combined to obtain the context information corresponding to the input understanding information.

[0160] Specifically, after the server obtains the input understanding information based on the large language model, the input understanding information is not directly sent to the terminal to be displayed to the user, but the private domain knowledge base and the network database (i.e. the Internet) are called to search the input understanding information respectively, and the first search information obtained based on the private domain knowledge base and the second search information obtained based on the network database are combined to obtain the context information corresponding to the input understanding information.

[0161] It should be noted that the present embodiment only describes one preferred embodiment of the present application, and in actual processing, the input understanding information can be searched in combination with the private domain knowledge base and / or the network database, and the present embodiment does not limit the search range and the search order.

[0162] In the present embodiment, by combining the private domain knowledge base containing rich teaching materials and the network database, the limitation of the general large language model in specific field knowledge is overcome, and it is ensured that the generated content not only conforms to the language logic, but also meets the specific requirements of the specific education institution, subject or teaching stage, thereby improving the professionalism and accuracy of the model generated content.

[0163] S603: The input understanding information and the context information are sent to the large language model to obtain an initial creative intention corresponding to the target creative information.

[0164] Optionally, after obtaining the input understanding information analyzed based on the target creative information of the user input and the context information analyzed and searched based on the input understanding information, the input understanding information and the context information are input to the large language model together, so that the server can combine the semantic processing capability of the large language model and the rich teaching materials in the private domain knowledge base to output more professional and accurate initial creative intention.

[0165] S604: After receiving the target creative intention, the target creative intention, the context information and the tool description information of each interactive tool in the interactive tool library are input to the large language model to obtain an initial creative outline corresponding to the target creative intention; the target creative intention is edited by the user based on the initial creative intention, and the interactive tool library includes a plurality of interactive capability tools for providing interactive information.

[0166] The interactive information includes but is not limited to ancient poems, Chinese character cards, mind maps, etc. The various interactive ability tools included in the interactive tool library can be described by their corresponding tool description information, which can specifically include a tool tag for identifying the interactive ability tool, and a function definition for describing the tool capability of the interactive ability tool. For example, the interactive ability tool for providing ancient poems can be identified by tags such as "ancient poems" and "poetry", and described by specific programming functions. The large language model can call the interactive ability tool according to the above "ancient poems" and "poetry" tags and specific programming functions, so that the generated initial creation outline contains the interactive ability tool for providing ancient poems, that is, the generated initial creation outline has the identification of the interactive ability tool for providing ancient poems after the outline node.

[0167] In an implementation, the collected various interactive ability tools with existing function definitions can be tagged, so as to form the function description and the corresponding tags of each interactive ability tool into the tool description information of each interactive ability tool, and build the interactive tool library.

[0168] It can be understood that in the process of obtaining the initial creation outline based on the target creation intention, the context information and the tool description information of the interactive tools in the interactive tool library, the large language model can first generate the content required to be presented on each page of the courseware according to the target creation intention and the context information, then match the content required to be presented on each page of the courseware with the tool description information of the various interactive ability tools in the interactive tool library, find one or more matching interactive ability tools, and output the initial creation outline with the identification of the interactive ability tool.

[0169] When the method is executed on the server side, the target creation intention received here can be the initial creation intention of step S603 sent to the client side, displayed by the client side, and the target creation intention obtained after the client side accepts the user's editing of the initial creation intention and sends the target creation intention to the server side. Thus, the server side receives the target creation intention. When the method is executed on the local client side, the initial creation intention can be directly displayed and the target creation intention can be obtained after receiving the user's editing result.

[0170] In this embodiment, by introducing various interactive ability tools in the interactive tool library, such as classroom activities, ancient poems, mind maps, word cards, etc., the attractiveness of the generated content can be significantly enhanced, the students' active thinking and in-depth exploration can be stimulated, and the interactivity of the courseware can be improved.

[0171] S605: After receiving the target creation outline, a courseware corresponding to the target creation outline is generated based on the target creation outline, the private knowledge base, the interactive tool library and the multi-modal material library. The target creation outline is edited by the user based on the initial creation outline, and the multi-modal material library includes courseware generation materials of multiple different modalities.

[0172] When the method is executed on the server side, the target creation outline received here can be the initial creation outline of step S604 sent to the client side, displayed by the client side, and the target creation outline obtained after the client side accepts the user's editing of the initial creation outline and sends the target creation outline to the server side. Thus, the server side receives the target creation outline. When the method is executed on the local client side, the initial creation outline can be directly displayed and the target creation outline can be obtained after receiving the editing result of the user.

[0173] Since the large language model directly generates a complete courseware and misses a lot of content due to information overload, the courseware generation task arrangement engine is introduced in the present application, and the courseware generation task is split into multiple sub-tasks according to chapters or courseware pages according to the courseware generation task arrangement engine and in combination with the target creation outline. Each sub-task calls the large language model for courseware generation, so that the granularity of the task handled by the large model is smaller and the quality of the generated content is higher.

[0174] Optionally, when generating the courseware corresponding to the target creation outline based on the target creation outline, the private knowledge base, the interactive tool library and the multi-modal material library, at least one instruction generation task corresponding to the target creation outline is first created according to the target creation outline. Different instruction generation tasks correspond to different outline nodes in the target creation outline. Then, based on the large language model, the context information, the interactive tool library and the multi-modal material library, the courseware generation instructions corresponding to each instruction generation task are obtained, and the courseware generation instructions corresponding to each instruction generation task are calibrated respectively. Finally, the interactive tool library and the multi-modal material library are called, and the courseware corresponding to the target creation outline is generated according to the calibrated courseware generation instructions corresponding to each instruction generation task.

[0175] Among them, the multi-modal material library includes courseware generation materials of multiple different modalities, specifically including materials of multiple formats such as text, pictures, audio, video, etc. Instruction calibration refers to correcting the text, adjusting the format or function format, filling in, etc. of the courseware generation instruction according to the preset adjustment rule or mapping table.

[0176] In this embodiment, by subdividing the courseware generation task into multiple instruction generation tasks, it can be ensured that each task focuses on a specific module, which helps to improve the quality of each part of the courseware; in addition, after obtaining the courseware generation instructions corresponding to each instruction generation task, the instruction calibration of each courseware generation instruction is performed, which can ensure the consistency and accuracy of each part of the courseware.

[0177] Specifically, when obtaining the courseware generation instructions corresponding to each instruction generation task based on the large language model, the context information, the interactive tool library and the multi-modal material library, the following information is input into the large language model: the outline text of the outline node corresponding to each instruction generation task, the context information, the tool description information of each interactive tool in the interactive tool library, the set of material tags corresponding to each courseware generation material in the multi-modal material library, and the set of preset layout formats of the courseware. Thus, the courseware generation instructions corresponding to each instruction generation task output by the large language model are obtained.

[0178] The set of preset layout formats of the courseware refers to a set of information of a series of layout specifications or style specifications that are preset to ensure that the generated courseware has a unified appearance and structure. For example, style information of the title and subtitle of the courseware, paragraph format information, layout format information of courseware generation materials of various different modalities such as pictures, etc. During the courseware generation process, these sets of preset layout formats of the courseware are used as one of the constraint conditions, and are input into the large language model together with the outline text of the outline node corresponding to each instruction generation task, the context information, the interactive tool description information and the material tags. In this way, the large model can consider the requirements of layout and style while generating courseware content, ensuring that the generated courseware meets both content requirements and appearance and structure standards.

[0179] In this embodiment, by inputting the outline text of the outline node corresponding to the instruction generation task into the large language model, it can be ensured that the generated courseware content is consistent with the outline content of the target creation outline. By inputting the context information into the large language model, the accuracy and professionalism of the courseware content can be ensured. By inputting the tool description information of each interactive tool in the interactive tool library into the large language model, the interactivity of the courseware content can be ensured. In addition, by inputting the set of preset layout formats of the courseware into the large language model, it can be ensured that the finally generated courseware meets the specifications and requirements of a specific format, and that the courseware meets the preset appearance and structure standards.

[0180] The courseware generation method can make the large language model better understand the professional knowledge in the preset field, and improve the pertinence and accuracy of the courseware content. In the process of generating the initial creation outline and the courseware, the interactive tool library containing a plurality of interactive ability tools is introduced to participate in the courseware design, which can effectively stimulate the active thinking and in-depth exploration of students, and enhance the interactivity of the courseware. In addition, the application also refers to the pre-constructed material library in the process of generating the courseware, and inputs a plurality of material tag sets corresponding to courseware generation materials in different modes into the large language model, so that the finally generated courseware content contains rich material information such as pictures, videos and classroom activity games, and the richness of the courseware content is enhanced. The courseware generated by the above method can ensure that the finally generated courseware has pertinence, accuracy, interactivity and richness, and accurately meets the actual teaching needs.

[0181] In one embodiment, as shown in FIG. 8, a method for constructing a private domain knowledge base is provided. In this embodiment, the method is applied to a server. It should be understood that the method can also be applied to a terminal, and can also be applied to a system including a terminal and a server, and can be realized through the interaction of the terminal and the server. In this embodiment, the method includes the following steps:

[0182] S801: dividing knowledge texts in a preset field according to a preset division rule to obtain at least two division segments.

[0183] The division of the knowledge texts in the preset field according to the preset division rule can be, but is not limited to, dividing teaching materials such as teacher's books and student's books according to knowledge units.

[0184] Optionally, in the process of constructing the private domain knowledge base, the knowledge in the education field is first abstractly divided. For example, the content in the teacher's book is divided into four division segments of teaching objectives, teaching material analysis, teaching suggestions and teaching design; and the content in the student's book is divided into three division segments of teaching material original text, character words and after-school thinking.

[0185] S802: performing text vectorization processing on the at least two division segments to obtain text vectorization data corresponding to the at least two division segments.

[0186] Optionally, before constructing the vector library based on the teaching materials such as the teacher's book and the student's book, the divided text segments are subjected to text vectorization, and the text is converted into vectorization data through a text vectorization model (such as an N-Gram model, a word embedding model, etc.).

[0187] S803: constructing a vector library according to the text vectorization data corresponding to the at least two division segments; the vector library is used for analyzing input understanding information.

[0188] Optionally, all the text-based student book vectorization data are collected together to build a student book vector library, and all the text-based teacher book vectorization data are collected together to build a teacher book vector library. The vector library is built based on the above-mentioned student book vector library and teacher book vector library, and is used for semantic analysis of the input understanding information.

[0189] S804: A business structural database is built according to the knowledge text of the preset field; the business structural database has a mapping relationship with the content in the vector library; and the business structural database is used for retrieval of the input understanding information.

[0190] It can be understood that the above-mentioned vector library has strong semantic analysis capability, but the retrieval capability of the above-mentioned vector library is poor because the complete teacher book content and student book content are split into multiple fragments according to the knowledge unit when the vector library is built. In order to avoid that the corresponding context information of the input understanding information is not complete and comprehensive, the embodiment further builds a teacher book structural database which is not divided and contains full-amount teacher book content and a student book structural database which is not divided and contains full-amount student book content. A business structural database is built based on the above-mentioned teacher book structural database and student book structural database, and is used for retrieval of the input understanding information.

[0191] It is worth noting that the student book vector library in the above-mentioned vector library has a corresponding relationship with the student book structural database in the above-mentioned business structural database, that is, when the corresponding context information of the input understanding information is obtained based on the private domain knowledge base, the input understanding information is first subjected to semantic analysis based on the student book vector library to obtain a target content fragment related to the input understanding information in the student book, and then the target content fragment is input into the student book structural database in the business structural database for retrieval to obtain complete context information related to the input understanding information. Similarly, the teacher book vector library in the above-mentioned vector library has a corresponding relationship with the teacher book structural database in the above-mentioned business structural database, that is, when the corresponding context information of the input understanding information is obtained based on the private domain knowledge base, the input understanding information is first subjected to semantic analysis based on the teacher book vector library to obtain a target content fragment related to the input understanding information in the teacher book, and then the target content fragment is input into the teacher book structural database in the business structural database for retrieval to obtain complete context information related to the input understanding information.

[0192] S805: A private domain knowledge base is built based on the vector library and the business structural database.

[0193] Optionally, a vector library is constructed based on the student textbook vector library and the teacher textbook vector library for semantic analysis of the input understanding information. A business structural database is constructed based on the teacher textbook structural database and the student textbook structural database for retrieval of the input understanding information. A private domain knowledge base is constructed based on the vector library and the business structural database for optimizing the output content of the large language model so that it can reference the preset field professional knowledge base outside the training data source before generating a response.

[0194] In the embodiment, the private domain knowledge base containing the vector library and the business structural database enables the preset field professional knowledge base to not only have strong semantic analysis capability but also have fast retrieval capability, thereby improving the professionalism and accuracy of the courseware.

[0195] In one embodiment, as shown in FIG. 9, a method for constructing a multi-modal material library is provided. The embodiment takes the method applied to a server as an example. It can be understood that the method can also be applied to a terminal and can also be applied to a system including a terminal and a server and is realized through the interaction of the terminal and the server. The method includes the following steps:

[0196] S901: Obtain courseware generation materials of multiple different modalities; the courseware generation materials of multiple different modalities include one or more of picture materials, video materials, and interactive materials.

[0197] Optionally, the courseware generation materials of multiple different modalities are obtained based on a historical courseware database and / or a text-to-image model. For example, the courseware of the historical courseware database can be parsed to extract pictures, videos, classroom activities, and other materials contained in the courseware historically created by the user, and then the content is audited to obtain the courseware generation materials of multiple different modalities. A batch of picture materials conforming to the textbook content can also be trained using the trained text-to-image model.

[0198] S902: The multi-modal large model is used to label the courseware generation materials of multiple different modalities to obtain material labels corresponding to the courseware generation materials of multiple different modalities.

[0199] Optionally, the multi-modal large model is used to label the courseware generation materials of multiple different modalities and / or the picture materials conforming to the textbook content to obtain multiple different dimension labels corresponding to the courseware generation materials of multiple different modalities. For example, material format labels, material type labels, and material usage rate labels.

[0200] S903: Construct a multi-modal material library according to the courseware generation materials of multiple different modalities and the material labels corresponding to the courseware generation materials of multiple different modalities.

[0201] Optionally, an auditing link of the courseware generation materials of the plurality of different modalities and the material tags corresponding to the courseware generation materials of the plurality of different modalities can be added, and after the accuracy is confirmed through manual auditing, the courseware generation materials of the plurality of different modalities and the material tags corresponding to the courseware generation materials of the plurality of different modalities are used to construct the multi-modal material library under the condition that the auditing is passed

[0202] In this embodiment, the courseware generation materials of the plurality of different modalities are obtained based on the historical courseware database and / or the text-to-image model, which effectively enriches the material content in the multi-modal material library. Through the multi-modal large model, accurate tags can be automatically added to the materials, which can greatly improve the organization and retrieval efficiency of the material library and shorten the courseware production cycle. In addition, when constructing the multi-modal material library, an auditing link for the courseware generation materials and the material tags corresponding to the courseware generation materials is added, so that it is ensured that the materials used for creating courseware do not have infringement problems.

[0203] Based on the inventive concept of the above courseware generation method, as shown in FIG. 10, the present embodiment further provides a courseware generation device 100 for implementing the above courseware generation method. The courseware generation device 100 comprises:

[0204] The first display module 101 is configured to display a courseware creation page, and the courseware creation page displays a first trigger control;

[0205] The receiving module 102 is configured to receive target creation information input by a user on the courseware creation page; the target creation information comprises creation demand information and / or a creation reference document;

[0206] The first response module 103 is configured to obtain an initial creation intention corresponding to the target creation information based on a large language model and a private domain knowledge base in response to a first trigger operation on the first trigger control; the private domain knowledge base is a knowledge base constructed by preprocessing knowledge texts of a preset field;

[0207] The second display module 104 is configured to display an intention editing page of the initial creation intention, and the intention editing page displays a second trigger control; the initial creation intention is editable;

[0208] The second response module 105 is configured to obtain an initial creation outline corresponding to the target creation intention based on a large language model, a private domain knowledge base and an interactive tool library after obtaining the target creation intention edited by a user based on the initial creation intention in response to a second trigger operation on the second trigger control; the interactive tool library comprises a plurality of interactive capability tools for providing interactive information;

[0209] The third display module 106 is configured to display a draft editing page of an initial creation draft, and the draft editing page displays a third trigger control; the initial creation draft includes a plurality of draft nodes, each draft node includes draft text, and at least one draft node displays an identifier of a corresponding interactive tool corresponding to the draft node; and the initial creation draft is editable.

[0210] The third response module 107 is configured to, in response to a third trigger operation on the third trigger control, obtain a target creation draft edited by a user based on the initial creation draft, and generate a courseware corresponding to the target creation draft based on a large language model, a private domain knowledge base, an interactive tool library, and a multi-modal material library; the multi-modal material library includes a plurality of courseware generation materials of different modalities.

[0211] The fourth display module 108 is configured to display a courseware display page including the courseware, and the courseware includes at least one interactive tool provided by the interactive tool library.

[0212] In an embodiment, the first response module 103 is specifically configured to input the target creation information into the large language model to obtain input understanding information; process the input understanding information based on the private domain knowledge base to obtain context information corresponding to the input understanding information; and send the input understanding information and the context information to the large language model to obtain an initial creation intent corresponding to the target creation information.

[0213] In an embodiment, the first response module 103 is specifically configured to analyze the creation reference document, and input reference document information and creation requirement information obtained by the analysis into the large language model to obtain input understanding information.

[0214] In an embodiment, the second response module 105 is specifically configured to input the target creation intent, the context information, and tool description information of each interactive tool in the interactive tool library into the large language model to obtain an initial creation draft corresponding to the target creation intent.

[0215] In an embodiment, the third response module 107 is specifically configured to create at least one instruction generation task corresponding to the target creation draft according to the target creation draft; different instruction generation tasks correspond to different draft nodes in the target creation draft; obtain courseware generation instructions corresponding to each instruction generation task based on the large language model, the context information, the interactive tool library, and the multi-modal material library; perform instruction calibration on the courseware generation instructions corresponding to each instruction generation task respectively; and call the interactive tool library and the multi-modal material library to generate the courseware corresponding to the target creation draft according to the calibrated courseware generation instructions corresponding to each instruction generation task.

[0216] In an embodiment, the third response module 107 is specifically configured to input the following information into the large language model: the outline text of the outline node corresponding to each instruction generation task, the context information, the tool description information of each interactive tool in the interactive tool library, the set of material tags corresponding to each courseware generation material in the multi-modal material library, and the set of preset layout formats of the courseware; and obtain the courseware generation instructions corresponding to each instruction generation task output by the large language model.

[0217] In an embodiment, the courseware generation apparatus 100 further comprises a private domain knowledge base construction module configured to divide the knowledge text of the preset domain according to a preset division rule to obtain at least two divided segments; perform text vectorization processing on the at least two divided segments to obtain text vectorization data corresponding to the at least two divided segments; construct a vector library according to the text vectorization data corresponding to the at least two divided segments; the vector library is configured to analyze the input understanding information; construct a business structural database according to the knowledge text of the preset domain; the business structural database has a mapping relationship with the content in the vector library; the business structural database is configured to search the input understanding information; and construct a private domain knowledge base based on the vector library and the business structural database.

[0218] In an embodiment, the courseware generation apparatus 100 further comprises a multi-modal material library construction module configured to obtain courseware generation materials of multiple different modalities; the courseware generation materials of multiple different modalities include one or more of picture materials, video materials, and interactive materials; perform tag processing on the courseware generation materials of multiple different modalities by using a multi-modal large model to obtain material tags corresponding to the courseware generation materials of multiple different modalities; and construct a multi-modal material library according to the courseware generation materials of multiple different modalities and the material tags corresponding to the courseware generation materials of multiple different modalities, in the case that the courseware generation materials of multiple different modalities pass the audit.

[0219] In an embodiment, the first response module 103 is specifically configured to search the input understanding information in the private domain knowledge base to obtain first search information; search the input understanding information through the Internet to obtain second search information; and combine the first search information and the second search information to obtain the context information corresponding to the input understanding information.

[0220] In an embodiment, the above-mentioned intention editing page further displays an intention editing control configured to receive an editing operation of the user on the initial creation intention.

[0221] In an embodiment, the above-mentioned initial creation outline comprises a plurality of end nodes, each end node is followed by a node editing control configured to receive an editing operation of the user on the outline text included in the end node.

[0222] In one embodiment, the above-mentioned outline editing page further displays at least two outline display form selection controls respectively for controlling the initial creation outline to be displayed in different display forms.

[0223] In one embodiment, the above-mentioned courseware display page includes a display area of the target creation outline and a preview area of the courseware; the target creation outline includes a plurality of outline nodes, and the outline text included in each outline node corresponds to the courseware content of the courseware; the courseware generation apparatus 100 further includes a fourth response module for receiving a fourth trigger operation of the user on the target outline node; the target outline node is any one of the plurality of outline nodes; in response to the fourth trigger operation, the courseware content corresponding to the outline text included in the target outline node is displayed in the preview area of the courseware.

[0224] Based on the above-mentioned application concept of the courseware generation method, as shown in FIG. 11, the present embodiment further provides a courseware generation apparatus 110 for implementing the above-mentioned courseware generation method applied to the server side. The courseware generation apparatus 110 includes:

[0225] The intention recognition module 111 is configured to input the target creation information into the large language model to obtain input understanding information; process the input understanding information based on the private domain knowledge base to obtain context information corresponding to the input understanding information; the private domain knowledge base is a knowledge base constructed by preprocessing knowledge texts of a preset domain, and the target creation information includes creation requirement information and / or creation reference documents; and send the input understanding information and the context information to the large language model to obtain an initial creation intention corresponding to the target creation information.

[0226] The outline generation module 112 is configured to input the target creation intention, the context information, and tool description information of each interactive tool in the interactive tool library into the large language model to obtain an initial creation outline corresponding to the target creation intention; the target creation intention is edited by the user based on the initial creation intention, and the interactive tool library includes a plurality of interactive capability tools for providing interactive information.

[0227] The courseware generation module 113 is configured to generate a courseware corresponding to the target creation outline based on the target creation outline, the private domain knowledge base, the interactive tool library, and a multi-modal material library; the target creation outline is edited by the user based on the initial creation outline, and the multi-modal material library includes a plurality of courseware generation materials of different modalities.

[0228] In an embodiment, the courseware generation module 113 is specifically configured to create at least one instruction generation task corresponding to the target creation outline according to the target creation outline; different instruction generation tasks correspond to different outline nodes in the target creation outline; obtain courseware generation instructions corresponding to each instruction generation task based on the large language model, the context information, the interactive tool library, and the multi-modal material library; respectively calibrate the courseware generation instructions corresponding to each instruction generation task; call the interactive tool library and the multi-modal material library, and generate the courseware corresponding to the target creation outline according to the calibrated courseware generation instructions corresponding to each instruction generation task.

[0229] In an embodiment, the courseware generation module 113 is specifically configured to input the following information into the large language model: the outline text of the outline node corresponding to each instruction generation task, the context information, the tool description information of each interactive tool in the interactive tool library, the set of material tags corresponding to each courseware generation material in the multi-modal material library, and the set of preset layout formats of the courseware; and obtain the courseware generation instructions corresponding to each instruction generation task output by the large language model.

[0230] In an embodiment, the intent recognition module 111 is specifically configured to parse the creation reference document, input the parsed reference document information and the creation demand information into the large language model, and obtain input understanding information.

[0231] In an embodiment, the courseware generation apparatus 110 further includes a private domain knowledge base construction module configured to divide the knowledge text of a preset domain according to a preset division rule to obtain at least two divided segments; perform text vectorization processing on the at least two divided segments to obtain text vectorization data corresponding to the at least two divided segments; construct a vector library according to the text vectorization data corresponding to the at least two divided segments; the vector library is used for analyzing the input understanding information; construct a business structural database according to the knowledge text of the preset domain; the business structural database has a mapping relationship with the content in the vector library; the business structural database is used for searching the input understanding information; and construct a private domain knowledge base based on the vector library and the business structural database.

[0232] In an embodiment, the courseware generation apparatus 110 further includes a multi-modal material library construction module configured to obtain a plurality of different modal courseware generation materials; the plurality of different modal courseware generation materials include one or more of picture class materials, video class materials, and interactive class materials; perform tag processing on the plurality of different modal courseware generation materials by using a multi-modal large model to obtain material tags corresponding to the plurality of different modal courseware generation materials; and construct a multi-modal material library according to the plurality of different modal courseware generation materials and the material tags corresponding to the plurality of different modal courseware generation materials.

[0233] The electronic device can also be a server, and an internal structure diagram of the electronic device can be as shown in FIG. 13. The electronic device includes a processor, a memory, an input / output interface (I / O), and a communication interface. The processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the electronic device is configured to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for running the operating system and the computer program in the non-volatile storage medium. The database of the electronic device is configured to store professional knowledge data in a preset field, multi-modal material data, and interactive ability tool data. The input / output interface of the electronic device is configured to exchange information between the processor and an external device. The communication interface of the electronic device is configured to communicate with an external terminal through a network connection. The processor of the electronic device executes the computer program to implement a courseware generation method.

[0234] It is worth noting that the electronic device can also be a server, and an internal structure diagram of the electronic device can be as shown in FIG. 13. The electronic device includes a processor, a memory, an input / output interface (I / O), and a communication interface. The processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the electronic device is configured to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for running the operating system and the computer program in the non-volatile storage medium. The database of the electronic device is configured to store professional knowledge data in a preset field, multi-modal material data, and interactive ability tool data. The input / output interface of the electronic device is configured to exchange information between the processor and an external device. The communication interface of the electronic device is configured to communicate with an external terminal through a network connection. The processor of the electronic device executes the computer program to implement a courseware generation method.

[0235] The embodiments of the present application further provide a computer readable storage medium, which stores instructions. When the instructions are executed on a computer or a processor, the computer or the processor performs one or more steps of the above-mentioned embodiments. When the components of the above-mentioned electronic device are realized in the form of software function units and sold or used as independent products, the components can be stored in the above-mentioned computer readable storage medium.

[0236] In the above embodiments, all or part of the embodiments can be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable apparatus. The computer instructions can be stored in or transmitted by a computer readable storage medium. The computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through a wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.) manner. The computer readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. integrated with one or more available media. The available medium can be a magnetic medium (such as a floppy disk, a hard disk, a magnetic tape), an optical medium (such as a digital versatile disc (DVD)) or a semiconductor medium (such as a solid state disk (SSD)) and the like.

[0237] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by a computer program to instruct related hardware, which can be stored in a computer readable storage medium. When the program is executed, it can include the processes of the above-mentioned embodiments. The storage medium includes ROM, RAM, magnetic or optical disk and other program code storage media. In the case of no conflict, the technical features in the embodiments and the embodiments can be combined arbitrarily.

[0238] The above-described embodiments are merely intended to describe the preferred embodiments of the present application, and are not intended to limit the scope of the present application. Various changes and modifications of the present application, which are apparent to those skilled in the art, can be made thereto without departing from the design spirit of the present application, and such changes and modifications should fall within the scope of the appended claims.

Claims

1. A method for generating a courseware, characterized by, The method comprises: receiving target creation information input by a user on the courseware creation page; the target creation information comprises creation requirement information and / or creation reference documents; obtaining an initial creation intention corresponding to the target creation information based on a large language model and a private domain knowledge base; the private domain knowledge base is a knowledge base constructed by preprocessing knowledge texts of a preset domain; displaying an intention editing page of the initial creation intention; after obtaining a target creation intention edited by the user based on the initial creation intention, obtaining an initial creation outline corresponding to the target creation intention based on the large language model, the private domain knowledge base and an interactive tool library; the interactive tool library comprises a plurality of interactive capability tools for providing interactive information; displaying an outline editing page of the initial creation outline; after obtaining a target creation outline edited by the user based on the initial creation outline, generating a courseware corresponding to the target creation outline based on the large language model, the private domain knowledge base, the interactive tool library and a multi-modal material library; displaying a courseware display page of the courseware; the courseware comprises at least one interactive capability tool provided by the interactive tool library.

2. The method of claim 1, wherein, The method comprises: inputting the target creation information into the large language model to obtain input understanding information; processing the input understanding information based on the private domain knowledge base to obtain context information corresponding to the input understanding information; sending the input understanding information and the context information to the large language model to obtain an initial creation intention corresponding to the target creation information.

3. The method of claim 2, wherein, In the case where the target creation information comprises the creation requirement information and the creation reference documents, the method comprises: parsing the creation reference documents, and inputting reference document information obtained by parsing and the creation requirement information into the large language model to obtain the input understanding information.

4. The method of claim 2, wherein, The method comprises: inputting the target creation intention, the context information and tool description information of each interactive tool of the interactive tool library into the large language model to obtain an initial creation outline corresponding to the target creation intention.

5. The method of claim 2, wherein, The method comprises: creating at least one instruction generation task corresponding to the target creation outline according to the target creation outline; different instruction generation tasks correspond to different outline nodes in the target creation outline; obtaining courseware generation instructions corresponding to each instruction generation task based on the large language model, the context information, the interactive tool library and the multi-modal material library; respectively performing instruction calibration on the courseware generation instructions corresponding to each instruction generation task; and Call the interactive tool library and the multi-modal material library, generate the calibrated courseware generation instruction corresponding to each instruction generation task according to the instructions, and generate the courseware corresponding to the target creation outline.

6. The method of claim 5, wherein, The courseware generation instruction corresponding to each instruction generation task is obtained based on the large language model, the context information, the interactive tool library and the multi-modal material library, comprising: Input the following information into the large language model: the outline text of the outline node corresponding to each instruction generation task, the context information, the tool description information of each interactive tool in the interactive tool library, the set of material tags corresponding to each courseware generation material in the multi-modal material library, and the set of preset layout formats of the courseware; Obtain the courseware generation instruction corresponding to each instruction generation task output by the large language model.

7. The method of claim 1, wherein, The private domain knowledge base is constructed in the following way: According to a preset division rule, the knowledge text of the preset field is divided to obtain at least two divided segments; Text vectorization processing is performed on the at least two divided segments to obtain text vectorization data corresponding to the at least two divided segments; A vector library is constructed according to the text vectorization data corresponding to the at least two divided segments; the vector library is used to analyze the input understanding information; A business structural database is constructed according to the knowledge text of the preset field; the business structural database has a mapping relationship with the content in the vector library; The business structural database is used to retrieve the input understanding information; The private domain knowledge base is constructed based on the vector library and the business structural database.

8. The method of claim 1, wherein, The multi-modal material library is constructed in the following way: Obtain a plurality of different modal courseware generation materials; the plurality of different modal courseware generation materials include one or more of picture class materials, video class materials and interactive class materials; A multi-modal large model is used to label the plurality of different modal courseware generation materials to obtain material tags corresponding to the plurality of different modal courseware generation materials; According to the plurality of different modal courseware generation materials and the material tags corresponding to the plurality of different modal courseware generation materials, the multi-modal material library is constructed.

9. The method of claim 2, wherein, The input understanding information is processed based on the private domain knowledge base to obtain context information corresponding to the input understanding information, comprising: In the private domain knowledge base, the input understanding information is retrieved to obtain first retrieval information; The input understanding information is retrieved through the Internet to obtain second retrieval information; The first retrieval information and the second retrieval information are combined to obtain the context information corresponding to the input understanding information.

10. The method of claim 1, wherein, The initial creation outline includes a plurality of end nodes, each end node corresponds to a node editing control after it, and the node editing control is used to receive the editing operation of the user on the outline text included in the end node.

11. The method of claim 1, wherein, The outline editing page also displays at least two outline display form selection controls, which are respectively used to control the initial creation outline to be displayed in different display forms.

12. The method of claim 1, wherein, The courseware display page includes a display area of the target creation outline and a preview area of the courseware; The target creation outline includes a plurality of outline nodes, and outline text included in each of the outline nodes corresponds to courseware content of the courseware. The method further includes: receiving a fourth trigger operation of the user on a target outline node; the target outline node being any one of the plurality of outline nodes; in response to the fourth trigger operation, displaying the courseware content corresponding to the outline text included in the target outline node in a preview area of the courseware.

13. The method of claim 12, wherein, The display area of the target creation outline includes the target creation outline and at least two outline display form selection controls, and the at least two outline display form selection controls are respectively used to control the target creation outline to be displayed in different display forms.

14. A method for generating a courseware, characterized by, The method includes: inputting target creation information into a large language model to obtain input understanding information; processing the input understanding information based on a private domain knowledge base to obtain context information corresponding to the input understanding information; the private domain knowledge base is a knowledge base constructed by preprocessing knowledge texts of a preset domain, and the target creation information includes creation requirement information and / or creation reference documents; sending the input understanding information and the context information to the large language model to obtain an initial creation intent corresponding to the target creation information; after receiving the target creation intent, inputting the target creation intent, the context information, and tool description information of each interactive tool in an interactive tool library into the large language model to obtain an initial creation outline corresponding to the target creation intent; the target creation intent is edited by a user based on the initial creation intent; after receiving the target creation outline, generating a courseware corresponding to the target creation outline based on the target creation outline, the private domain knowledge base, the interactive tool library, and a multi-modal material library; the target creation outline is edited by the user based on the initial creation outline.

15. The method of claim 14, wherein, The generating of the courseware corresponding to the target creation outline based on the target creation outline, the private domain knowledge base, the interactive tool library, and the multi-modal material library includes: creating at least one instruction generation task corresponding to the target creation outline according to the target creation outline; different instruction generation tasks correspond to different outline nodes in the target creation outline; obtaining courseware generation instructions corresponding to each instruction generation task based on the large language model, the context information, the interactive tool library, and the multi-modal material library; performing instruction calibration on the courseware generation instructions corresponding to each instruction generation task respectively; calling the interactive tool library and the multi-modal material library to generate the courseware corresponding to the target creation outline according to the calibrated courseware generation instructions corresponding to each instruction generation task.

16. The method of claim 15, wherein, The obtaining of the courseware generation instructions corresponding to each instruction generation task based on the large language model, the context information, the interactive tool library, and the multi-modal material library includes: The large language model is input with the following information: the outline text of the outline node corresponding to each instruction generation task, the context information, the tool description information of each interactive tool in the interactive tool library, the set of material tags corresponding to each courseware generation material in the multi-modal material library, and the set of layout formats of the preset courseware; Obtain the courseware generation instructions corresponding to each instruction generation task output by the large language model.

17. The method of claim 14, wherein, In the case where the target creation information includes the creation requirement information and the creation reference document, the input of the target creation information into the large language model to obtain the input understanding information includes: Parsing the creation reference document, and inputting the parsed reference document information and the creation requirement information into the large language model to obtain the input understanding information.

18. The method of claim 14, wherein, The private domain knowledge base is constructed in the following way: Divide the knowledge text of the preset field according to a preset division rule to obtain at least two divided segments; Text vectorization processing is performed on the at least two divided segments to obtain text vectorization data corresponding to the at least two divided segments; A vector library is constructed according to the text vectorization data corresponding to the at least two divided segments; the vector library is used for analyzing the input understanding information; A business structural database is constructed according to the knowledge text of the preset field; the business structural database has a mapping relationship with the content in the vector library; The business structural database is used for searching the input understanding information; The private domain knowledge base is constructed based on the vector library and the business structural database.

19. The method of claim 14, wherein, The multi-modal material library is constructed in the following way: Obtain courseware generation materials of multiple different modalities; the courseware generation materials of multiple different modalities include one or more of picture materials, video materials, and interactive materials; A multi-modal large model is used to label the courseware generation materials of multiple different modalities to obtain material tags corresponding to the courseware generation materials of multiple different modalities; The multi-modal material library is constructed according to the courseware generation materials of multiple different modalities and the material tags corresponding to the courseware generation materials of multiple different modalities.

20. The method of claim 14, wherein, The input understanding information is processed based on the private domain knowledge base to obtain context information corresponding to the input understanding information, which includes: In the private domain knowledge base, the input understanding information is searched to obtain first search information; The input understanding information is searched through the Internet to obtain second search information; The first search information and the second search information are combined to obtain the context information corresponding to the input understanding information.

21. A method for generating a courseware, characterized by The method includes: Displaying a courseware creation page; Receiving target creation information input by a user on the courseware creation page; the target creation information includes creation requirement information and / or a creation reference document; Displaying an intent editing page, the intent editing page displaying an initial creation intent corresponding to the target creation information; the initial creation intent is obtained based on a large language model and a private domain knowledge base; Receiving a first editing operation of the user on the initial creation intent; displaying a courseware display page, the courseware display interface displaying a courseware corresponding to the target courseware outline, the target courseware outline being obtained by editing the initial courseware outline according to the second editing operation, and the courseware being obtained based on the large language model, the private domain knowledge base, the interactive tool library, and the multi-modal material library, and the courseware including at least one interactive capability tool provided by the interactive tool library. receiving a second editing operation of the user on the initial courseware outline; displaying a courseware display page, the courseware display interface displaying a courseware corresponding to the target courseware outline, the target courseware outline being obtained by editing the initial courseware outline according to the second editing operation, and the courseware being obtained based on the large language model, the private domain knowledge base, the interactive tool library, and the multi-modal material library, and the courseware including at least one interactive capability tool provided by the interactive tool library. The initial creation intention is obtained by:

22. The method of claim 21, wherein, inputting the target creation information into the large language model to obtain input understanding information; processing the input understanding information based on the private domain knowledge base to obtain context information corresponding to the input understanding information; sending the input understanding information and the context information to the large language model to obtain the initial creation intention corresponding to the target creation information. In the case where the target creation information includes the creation requirement information and the creation reference document, the inputting the target creation information into the large language model to obtain input understanding information includes:

23. The method of claim 22, wherein, parsing the creation reference document, and inputting the parsed reference document information and the creation requirement information into the large language model to obtain the input understanding information. The initial courseware outline corresponding to the target creation intention is obtained based on the large language model, the private domain knowledge base, and the interactive tool library, and includes:

24. The method of claim 22, wherein, inputting the target creation intention, the context information, and tool description information of each interactive tool of the interactive tool library into the large language model to obtain the initial courseware outline corresponding to the target creation intention. The courseware is obtained by:

25. The method of claim 22, wherein, creating at least one instruction generation task corresponding to the target courseware outline according to the target courseware outline, different instruction generation tasks corresponding to different outline nodes in the target courseware outline; obtaining courseware generation instructions corresponding to each instruction generation task based on the large language model, the context information, the interactive tool library, and the multi-modal material library; respectively calibrating the courseware generation instructions corresponding to each instruction generation task; calling the interactive tool library and the multi-modal material library, and generating the courseware corresponding to the target courseware outline according to the calibrated courseware generation instructions corresponding to each instruction generation task. The courseware generation instructions corresponding to each instruction generation task are obtained based on the large language model, the context information, the interactive tool library, and the multi-modal material library, and include:

26. The method of claim 25, wherein, ​ The following information is input into the large language model: the outline text of each outline node corresponding to the instruction generation task, the context information, the tool description information of each interactive tool in the interactive tool library, the set of material tags corresponding to each courseware generation material in the multi-modal material library, and the set of layout formats of the preset courseware; Obtain the courseware generation instructions corresponding to each instruction generation task output by the large language model.

27. The method of claim 21, wherein, The private domain knowledge base is constructed in the following way: According to the preset division rule, the knowledge text of the preset field is divided to obtain at least two divided segments; Text vectorization processing is performed on the at least two divided segments to obtain text vectorization data corresponding to the at least two divided segments; According to the text vectorization data corresponding to the at least two divided segments, a vector library is constructed; the vector library is used to analyze the input understanding information; According to the knowledge text of the preset field, a business structural database is constructed; the business structural database has a mapping relationship with the content in the vector library; The business structural database is used to retrieve the input understanding information; Based on the vector library and the business structural database, the private domain knowledge base is constructed. The multi-modal material library is constructed in the following way:

28. The method of claim 21, wherein, Obtain a plurality of different modal courseware generation materials; the plurality of different modal courseware generation materials include one or more of picture materials, video materials and interactive materials; A multi-modal large model is used to label the plurality of different modal courseware generation materials to obtain material tags corresponding to the plurality of different modal courseware generation materials; According to the plurality of different modal courseware generation materials and the material tags corresponding to the plurality of different modal courseware generation materials, the multi-modal material library is constructed. The input understanding information is processed based on the private domain knowledge base to obtain context information corresponding to the input understanding information, including:

29. The method of claim 22, wherein, In the private domain knowledge base, the input understanding information is retrieved to obtain first retrieval information; The input understanding information is retrieved through the Internet to obtain second retrieval information; The first retrieval information and the second retrieval information are combined to obtain the context information corresponding to the input understanding information. The initial creation outline includes a plurality of end nodes, each end node corresponds to a node editing control, and the node editing control is used to receive the editing operation of the user on the outline text included in the end node.

30. The method of claim 21, wherein, The outline editing page also displays at least two outline display form selection controls, which are respectively used to control the initial creation outline to be displayed in different display forms.

31. The method of claim 21, wherein, The courseware display page includes a display area of the target creation outline and a preview area of the courseware; 32. The method of claim 21, wherein, The target creation outline includes a plurality of outline nodes, and the outline text included in each outline node corresponds to the courseware content of the courseware; The method further includes: Receiving a fourth trigger operation of the user on a target outline node; the target outline node is any one of the plurality of outline nodes; ​ In response to the fourth trigger operation, displaying, in a preview area of the courseware, courseware content corresponding to the outline text included in the target outline node.

33. The method of claim 32, wherein, The display area of the target creation outline includes the target creation outline and at least two outline display form selection controls, and the at least two outline display form selection controls are respectively used to control the target creation outline to be displayed in different display forms.

34. An electronic device, comprising: Comprise: A processor and a memory; the memory stores a computer program, and the processor implements the method steps of any one of claims 1-13, 14-20 or 21-33 when executing the computer program.

35. A computer storage medium, comprising: The computer storage medium stores a plurality of instructions, and the instructions are suitable for being loaded and executed by the processor as the method steps of any one of claims 1-13, 14-20 or 21-33.

Images