Content generation method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention discloses a content generation method and device, electronic equipment and a storage medium, and the method comprises the steps: running a content generation service which logs in a target account; obtaining first prompt information input to the content generation service, and when a knowledge base-based mode is selected in the content generation service for content generation, generating target content corresponding to the first prompt information in the content generation service based on a knowledge base pre-created by the target account; according to the embodiment, the accuracy of the target content can be effectively improved.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to a content generation method, device, electronic device and storage medium. Background Art

[0002] With the development of artificial intelligence technology, the application of content generation based on artificial intelligence has become more and more widespread. At present, when generating content, the network search is often carried out according to the prompt information input by the user, and then the model inference is carried out according to the search results to obtain the content generation results. Since the information searched on the network is generally general information, or there may be erroneous information, the accuracy of the content generation results is reduced. Summary of the invention

[0003] The following is a summary of the subject matter of the detailed description of the present disclosure. This summary is not intended to limit the scope of the claims.

[0004] The embodiments of the present disclosure provide a content generation method, device, electronic device and storage medium, which can effectively improve the accuracy of target content.

[0005] In one aspect, an embodiment of the present disclosure provides a content generation method, comprising:

[0006] Running a content generation service, wherein the content generation service is logged into a target account;

[0007] A first prompt information input to the content generation service is obtained, and when a knowledge base-based method is selected in the content generation service to generate content, target content corresponding to the first prompt information is generated in the content generation service based on the knowledge base pre-created by the target account.

[0008] On the other hand, the present disclosure also provides a content generation device, including:

[0009] An acquisition module, used to run a content generation service, wherein the content generation service is logged in with a target account;

[0010] A generation module is used to obtain first prompt information input into the content generation service, and when a knowledge base-based method is selected in the content generation service to generate content, the target content corresponding to the first prompt information is generated in the content generation service based on the knowledge base pre-created by the target account.

[0011] Furthermore, the first prompt information is input through a session page of the content generation service, and the session page displays a generation mode selection control, and the generation module is specifically used for:

[0012] When the knowledge base-based method is selected to generate content through the generation method selection control, the target content corresponding to the first prompt information is generated in the session page based on the knowledge base pre-created by the target account.

[0013] Furthermore, the above generation module is also used for:

[0014] displaying at least one first reference link in the session page;

[0015] The first reference link is used to access reference content in the knowledge base used to generate the target content.

[0016] Furthermore, the above-mentioned content generation device further includes a regeneration module, and the regeneration module is specifically used for:

[0017] In response to an operation of selecting at least one of the first reference links in the conversation page, marking the selected first reference link in the conversation page;

[0018] In response to the operation of regenerating the target content in the conversation page, the target content is regenerated in the conversation page based on the reference content corresponding to the selected first reference link.

[0019] Furthermore, the above generation module is specifically used for:

[0020] When the knowledge base-based method is selected to generate content through the generation method selection control, a file import control is displayed;

[0021] In response to the operation of importing a temporary file through the file import control, target content corresponding to the first prompt information is generated in the session page based on the knowledge base pre-created by the target account and the temporary file.

[0022] Furthermore, the above generation module is specifically used for:

[0023] When the knowledge base-based method is selected for content generation through the generation method selection control, a knowledge base page is displayed, wherein the knowledge base page is used to display reference content in the knowledge base pre-created by the target account;

[0024] In response to an operation of selecting at least one reference content in the knowledge base page, a target content corresponding to the first prompt information is generated in the conversation page based on the selected reference content.

[0025] Furthermore, the above content generation device further includes a first updating module, and the first updating module is specifically used for:

[0026] In response to an operation of adding content to the knowledge base, the reference content browsed or processed by the target account is added to the knowledge base.

[0027] Furthermore, the first update module is specifically used for:

[0028] Displaying a content import page of the knowledge base;

[0029] In response to an operation of adding content to the knowledge base in the content import page, the reference content imported by the target account is added to the knowledge base.

[0030] Furthermore, the first update module is specifically used for:

[0031] Displaying a conversation page of the content generation service, and acquiring second prompt information input in the conversation page;

[0032] When generating content in the conversation page, generating reference content corresponding to the second prompt information in the conversation page;

[0033] In response to an operation of adding content to the knowledge base in the conversation page, the reference content generated by the target account through the conversation is added to the knowledge base.

[0034] Furthermore, the first update module is specifically used for:

[0035] In response to an operation of adding content to the knowledge base in the conversation page, displaying a new note control;

[0036] In response to the operation of triggering the new note control, displaying a note page, wherein the note page includes a note list subpage and a note editing subpage displayed in parallel, and the note editing subpage displays a first adding control;

[0037] Highlighting the target note of the new note control operation in the note list subpage, and displaying the reference content generated by the target account through the conversation in the note editing subpage corresponding to the target note;

[0038] The reference content is edited in the note editing subpage, and in response to triggering the operation of the first adding control, the edited reference content is added to the knowledge base.

[0039] Furthermore, the first update module is specifically used for:

[0040] In response to selecting at least a portion of the sub-content of the reference content in the note editing sub-page, displaying an editing control, wherein the editing control is used to input an editing requirement or to trigger a preset editing function;

[0041] In response to the operation of triggering the editing control, generating editing content and a first insert control for inserting the editing content in the note editing subpage, and highlighting the editing content;

[0042] In response to the operation of triggering the first insert control, the edited content is added to the reference content, and the highlighting of the edited content is canceled.

[0043] Furthermore, the first update module is specifically used for:

[0044] In response to selecting at least a portion of the sub-content of the reference content in the note editing sub-page, displaying an interpretation control;

[0045] In response to the operation of triggering the interpretation control, displaying an interpretation page, displaying the interpretation content of the sub-content and a second adding control in the interpretation page;

[0046] In response to the operation of triggering the second adding control, adding the interpretation content and a second inserting control for inserting the interpretation content in the note editing subpage, and highlighting the interpretation content;

[0047] In response to the operation of triggering the second insertion control, the interpretation content is added to the reference content, and the highlighting of the interpretation content is canceled.

[0048] Furthermore, the first update module is specifically used for:

[0049] Displaying the media pages browsed by the target account;

[0050] In response to an operation of adding content to the knowledge base in the media page, the reference content in the media page is added to the knowledge base.

[0051] Furthermore, the first update module is specifically used for:

[0052] Displaying a session page of the content generation service, and when a network search-based method is selected in the session page to generate content, displaying at least one second reference link after generating content in the session page;

[0053] In response to the operation of opening the second reference link, a media page browsed by the target account is displayed.

[0054] Furthermore, the first update module is specifically used for:

[0055] In response to an operation of adding content to the knowledge base, obtaining media browsed or processed by the target account;

[0056] The medium is parsed to obtain reference content, and the reference content is added to the knowledge base.

[0057] Furthermore, the first update module is also used for:

[0058] During the process of parsing the medium, the parsing progress of the medium is displayed on a knowledge base page of the knowledge base.

[0059] Furthermore, the medium is a web page, and the first updating module is specifically used for:

[0060] Parsing the body and title of the webpage to obtain a first parsing result;

[0061] When it is determined that the first parsing result includes a path mark, the first parsing result is parsed again for the body and title to obtain a second parsing result;

[0062] When it is detected that no content exists in the second parsing result, extracting a text node in the web page as the second parsing result by using a path rule;

[0063] The second analysis result is cleaned of abnormal data to obtain reference content.

[0064] Furthermore, the above-mentioned content generation device further includes a display module, and the display module is specifically used for:

[0065] Displaying a knowledge base page of the knowledge base, wherein the knowledge base page displays at least one third reference link, and the third reference link is used to access the reference content in the knowledge base;

[0066] In response to an operation of opening the third reference link, the corresponding reference content is displayed.

[0067] Furthermore, the content generation device further includes a second updating module, which is specifically configured to:

[0068] In response to an operation of modifying the currently displayed reference content, updating the modified reference content in the knowledge base;

[0069] In response to an operation of reopening the third reference link, the updated reference content is displayed.

[0070] On the other hand, an embodiment of the present disclosure further provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the above-mentioned content generation method when executing the computer program.

[0071] On the other hand, an embodiment of the present disclosure further provides a computer-readable storage medium, wherein the storage medium stores a computer program, and the computer program is executed by a processor to implement the above-mentioned content generation method.

[0072] On the other hand, the embodiment of the present disclosure further provides a computer program product, which includes a computer program, and the computer program is stored in a computer-readable storage medium. The processor of the computer device reads the computer program from the computer-readable storage medium, and the processor executes the computer program, so that the computer device executes the above-mentioned content generation method.

[0073] The disclosed embodiments include at least the following beneficial effects: by running a content generation service logged in to a target account, first prompt information input into the content generation service is obtained; when a knowledge base-based method is selected in the content generation service to generate content, target content corresponding to the first prompt information is generated in the content generation service based on the knowledge base; since the knowledge base is pre-created by the target account, the target account can store private professional knowledge through the knowledge base, and based on the knowledge base, target content that is more refined and more in line with the target account's private knowledge field can be generated, making the target content more focused and professional, thereby effectively improving the accuracy of the target content.

[0074] Other features and advantages of the present disclosure will be set forth in the following description, and in part will be apparent from the description, or may be learned by practicing the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] The accompanying drawings are used to provide further understanding of the technical solution of the present disclosure and constitute a part of the specification. Together with the embodiments of the present disclosure, they are used to explain the technical solution of the present disclosure and do not constitute a limitation on the technical solution of the present disclosure.

[0076] Figure 1 A schematic diagram of an optional implementation environment provided for an embodiment of the present disclosure;

[0077] Figure 2 A schematic diagram of an optional flow chart of a content generation method provided in an embodiment of the present disclosure;

[0078] Figure 3 A schematic diagram of an optional operation of a conversation page when asking questions in the first question-and-answer scenario provided in an embodiment of the present disclosure;

[0079] Figure 4 A schematic diagram of an optional operation of a conversation page when answering in the first question-and-answer scenario provided by an embodiment of the present disclosure;

[0080] Figure 5An optional operation diagram of an initial conversation page in a writing scenario provided by an embodiment of the present disclosure;

[0081] Figure 6 An optional operation diagram of a conversation page when inputting writing requirements in a writing scenario provided by an embodiment of the present disclosure;

[0082] Figure 7 An optional operation diagram of a conversation page for a preliminary reply in a writing scenario provided by an embodiment of the present disclosure;

[0083] Figure 8 An optional operation diagram of a conversation page for further replying in a writing scenario provided by an embodiment of the present disclosure;

[0084] Fig. 9 A schematic diagram of an optional operation of a conversation page when asking questions in the second question-and-answer scenario provided by an embodiment of the present disclosure;

[0085] Fig.10 An optional operation diagram of a conversation page for the first reply in the second question-and-answer scenario provided by an embodiment of the present disclosure;

[0086] Fig.11 A schematic diagram of an optional operation of a conversation page when replying again in the second question-and-answer scenario provided by an embodiment of the present disclosure;

[0087] Fig.12 An optional operation diagram of a knowledge base page in a second conversation scenario provided by an embodiment of the present disclosure;

[0088] Fig.13 An optional operation diagram of a content import page provided in an embodiment of the present disclosure;

[0089] Fig.14 Another optional operation diagram of the content import page provided in the embodiment of the present disclosure;

[0090] Fig.15 An optional operation diagram of a media page provided in an embodiment of the present disclosure;

[0091] Fig.16 Another optional operation diagram of the media page provided in the embodiment of the present disclosure;

[0092] Fig.17 An optional schematic diagram showing parsing progress provided in an embodiment of the present disclosure;

[0093] Fig.18 , Fig.19 , Fig. 20 , Fig.21A schematic diagram of an optional process for parsing a web page provided in an embodiment of the present disclosure;

[0094] Fig. 22 An optional operation diagram for displaying reference content provided in an embodiment of the present disclosure;

[0095] Fig.23 An optional schematic diagram of updating reference content provided in an embodiment of the present disclosure;

[0096] Fig.24 An optional schematic diagram of knowledge base reconstruction and refreshing provided by an embodiment of the present disclosure;

[0097] Fig.25 An optional operation diagram of generating target content provided by an embodiment of the present disclosure;

[0098] Fig.26 A schematic diagram of the overall architecture of the intelligent workbench provided in an embodiment of the present disclosure;

[0099] Fig. 27 A schematic diagram of an optional architecture for knowledge base retrieval provided in an embodiment of the present disclosure;

[0100] Fig.28 An optional schematic diagram of a quantization result provided by an embodiment of the present disclosure;

[0101] Fig.29 An optional schematic diagram of an organizational structure provided in an embodiment of the present disclosure;

[0102] Fig.30 An optional schematic diagram of a search process provided by an embodiment of the present disclosure;

[0103] Fig.31 An optional flow chart of outputting target content provided in an embodiment of the present disclosure;

[0104] Fig.32 A schematic diagram of an operation of initiating editing of generated target content provided by an embodiment of the present disclosure;

[0105] Fig.33 An operational schematic diagram of an editing process of generated target content provided by an embodiment of the present disclosure;

[0106] Fig.34 An operational diagram of a process of inserting and editing content provided by an embodiment of the present disclosure;

[0107] Fig.35 An operational schematic diagram of an interpretation process provided by an embodiment of the present disclosure;

[0108] Fig.36A schematic diagram of an optional interface of the intelligent workbench provided in an embodiment of the present disclosure;

[0109] Fig.37 A schematic diagram of an optional structure of a content generation device provided in an embodiment of the present disclosure;

[0110] Fig.38 A partial structural block diagram of a terminal provided in an embodiment of the present disclosure;

[0111] Fig.39 A partial structural block diagram of a server provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0112] In order to make the purpose, technical solution and advantages of the present disclosure more clear, the present disclosure is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present disclosure and are not used to limit the present disclosure.

[0113] It should be noted that in various specific embodiments of the present disclosure, when it comes to the need to perform relevant processing based on data related to the characteristics of the target object such as the target object attribute information or attribute information set, the permission or consent of the target object will be obtained first, and the collection, use and processing of these data will comply with relevant laws, regulations and standards. Among them, the target object can be a user. In addition, when the embodiment of the present disclosure needs to obtain the attribute information of the target object, the separate permission or separate consent of the target object will be obtained through a pop-up window or by jumping to a confirmation page. After clearly obtaining the separate permission or separate consent of the target object, the necessary target object-related data used to enable the normal operation of the embodiment of the present disclosure will be obtained.

[0114] In the embodiments of the present disclosure, the term "module" or "unit" refers to a computer program or a part of a computer program with a predetermined function, and works together with other related parts to achieve a predetermined goal, and can be implemented in whole or in part by using software, hardware (such as processing circuits or memories) or a combination thereof. Similarly, a processor (or multiple processors or memories) can be used to implement one or more modules or units. In addition, each module or unit can be part of an overall module or unit that includes the function of the module or unit.

[0115] With the development of artificial intelligence technology, the application of content generation based on artificial intelligence has become more and more widespread. At present, when generating content, the network search is often carried out according to the prompt information input by the user, and then the model inference is carried out according to the search results to obtain the content generation results. Since the information searched on the network is generally general information, or there may be erroneous information, the accuracy of the content generation results is reduced.

[0116] Based on this, the embodiments of the present disclosure provide a content generation method, device, electronic device and storage medium, which can effectively improve the accuracy of target content.

[0117] Reference Figure 1 , Figure 1 A schematic diagram of an optional implementation environment provided for an embodiment of the present disclosure, the implementation environment includes a terminal 101 and a server 102, wherein the terminal 101 and the server 102 are connected via a communication network.

[0118] Exemplarily, terminal 101 can run a content generation service, wherein the content generation service is logged in to a target account; obtain first prompt information input into the content generation service, and when a knowledge base-based method is selected in the content generation service to generate content, the first prompt information is sent to server 102, and server 102 generates target content based on the knowledge base pre-created by the target account. Server 102 sends the target content to terminal 101, so that terminal 101 generates the target content corresponding to the first prompt information in the content generation service.

[0119] Terminal 101 obtains the first prompt information input into the content generation service by running the content generation service logged in to the target account. When the knowledge base-based method is selected to generate content in the content generation service, the target content corresponding to the first prompt information is generated in the content generation service based on the knowledge base. Since the knowledge base is pre-created by the target account, the target account can store private professional knowledge through the knowledge base. Based on the knowledge base, it can generate more refined target content that is more in line with the private knowledge field of the target account, making the target content more focused and professional, thereby effectively improving the accuracy of the target content.

[0120] Server 102 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms. In addition, server 102 can also be a node server in a blockchain network.

[0121] The terminal 101 may be a mobile phone, a computer, an intelligent voice interaction device, an intelligent home appliance, a vehicle-mounted terminal, etc., but is not limited thereto. The terminal 101 and the server 102 may be directly or indirectly connected via wired or wireless communication, which is not limited in the embodiments of the present disclosure.

[0122] Reference Figure 2 , Figure 2An optional flow chart of a content generation method provided in an embodiment of the present disclosure, the content generation method can be executed by a terminal, or can also be executed by a server in cooperation with a terminal, and the content generation method includes but is not limited to the following steps 201 to 202.

[0123] Step 201: Run the content generation service.

[0124] Among them, the content generation service refers to a service that can generate content. The content generated by the content generation service can be in the form of one or more of text, image, audio and video. The generated content can also be in other forms, which are not limited in the embodiments of the present disclosure. The content generation service can be in the form of an application, a web page, a browser plug-in or an API interface, etc. For example, a terminal can use the content generation service by installing and running a corresponding client. The client can be a proprietary client of the content generation service, or a web client or a client of at least one of the following applications: a social application, an information application, a reader, a shopping application, and a video application.

[0125] Among them, the content generation service is logged in with a target account, which is equivalent to the target account having been successfully logged in. The target account is a unique identity identifier used to log in to and access the content generation service. The target account is usually generated through user registration or through authorization on a third-party platform. For example, the target account can be an email address, user name or mobile phone number, etc. The target account can also be an account logged in through a third-party platform such as a social media or corporate account.

[0126] Taking the server cooperating with the terminal to execute the content generation method as an example, the server usually includes a user account database, which is used to store data of multiple user accounts, such as user account avatars, user account nicknames, etc. The target account is one of the multiple user accounts. In addition, the target account can allow users to log in to the same content generation service across terminals. For example, users can use the same target account to access the content generation service on a mobile phone, tablet or computer.

[0127] In a possible implementation, the content generation service is run, specifically by displaying a login page, where the user can log in to a target account. When the target account is successfully logged in, the content generation service is run.

[0128] Based on this, the target account is used to verify the user's identity, for example, through the target account's account password, mobile phone SMS verification code, email verification code or third-party platform authorization. The content generation service will only run after the target account has successfully logged in, ensuring that the user can safely access the content generation service, content generation records and personalized configurations, etc., which can prevent unauthorized access and avoid data leakage, thereby ensuring the security of user data and generated content.

[0129] Step 202: Obtain first prompt information input to the content generation service. When a knowledge base-based method is selected in the content generation service to generate content, target content corresponding to the first prompt information is generated in the content generation service based on the knowledge base pre-created by the target account.

[0130] Among them, the first prompt information is used to guide the content generation service to generate content. The first prompt information usually contains a specific intention. For example, the first prompt information usually contains the main goal, problem or requirement. When generating content based on the knowledge base, it can effectively identify and meet the intention contained in the first prompt information, so that the target content can effectively respond to the core needs in the first prompt information.

[0131] Specifically, the first prompt information may be in a form of one or more of text, image, audio, and video. The first prompt information may also be in other forms, which are not limited in the embodiments of the present disclosure.

[0132] Among them, the knowledge base is a database used for knowledge management, which can be specifically regarded as a collection of knowledge. By creating a knowledge base, it is helpful to collect, organize and extract knowledge in a specific field. The knowledge base provides knowledge support related to the received information for the content generation service. This knowledge can include basic facts, rules and other content. Optionally, the knowledge base can store knowledge in various forms, such as key-value pairs, knowledge graphs, feature vectors, etc. This structured storage method of knowledge can effectively improve the organization and utilization efficiency of knowledge.

[0133] It should be noted that the knowledge base is pre-created by the target account, so that there is an association between the knowledge base and the target account, and the knowledge in the knowledge base is private to the target account, which is equivalent to the knowledge base being created based on the private data of the target account. For example, when the target account is generated, an empty knowledge base is created and associated with the target account, and then the content browsed or processed by the target account can be added to the knowledge base of the target account, or the knowledge base shared by other accounts can be added to the knowledge base of the target account. Taking the shared knowledge base as an example, the user can add all or part of the knowledge in the shared knowledge base to the knowledge base of the target account. Among them, the sharing method of the knowledge base can be divided into encrypted sharing and public sharing. Encrypted sharing means that a password is attached when generating a link to the shared knowledge base. Only users who have both the link and the password can access the shared knowledge base. Public sharing means that no password is required when generating a link to the shared knowledge base. Any user who has the link can access the shared knowledge base.

[0134] Among them, choosing a knowledge base-based method for content generation specifically means that the generation process of the target content needs to refer to the knowledge in the knowledge base, so that the target content can include the basic facts, rules and other contents contained in the knowledge in the knowledge base; in addition, the referenced knowledge in the knowledge base can be knowledge related to the first prompt information, ensuring that the generated target content not only has reference value, but also can respond to the core needs in the first prompt information.

[0135] Based on this, by running the content generation service logged in to the target account, the first prompt information input into the content generation service is obtained. When the knowledge base-based method is selected to generate content in the content generation service, the target content corresponding to the first prompt information is generated in the content generation service based on the knowledge base. Since the knowledge base is pre-created by the target account, the target account can store private professional knowledge through the knowledge base. Based on the knowledge base, it can generate more refined target content that is more in line with the private knowledge field of the target account, making the target content more focused and professional, thereby effectively improving the accuracy of the target content.

[0136] Specifically, content generation is performed based on a knowledge base pre-created by a target account. Specifically, content generation may be performed based on knowledge within a specified range in the knowledge base. The specified range may be preset or user-defined. For example, the specified range may be all knowledge or part of the knowledge in the knowledge base. By setting the specified range, the flexibility of the knowledge referenced for content generation can be improved, which helps to generate high-quality target content.

[0137] In addition, assuming that the knowledge base contains multiple reference contents, you can also first determine the correlation score between each reference content in the knowledge base and the first prompt information, and then sort the reference contents in descending order based on the correlation score, and then generate content for the reference contents in the top N positions, where N is a positive integer. For example, N can be 1. It can automatically select reference knowledge with a higher correlation score in the knowledge base for content generation without the need for users to select by themselves, which can reduce time consumption and effectively improve the generation efficiency of the target content while ensuring the generation quality of the target content.

[0138] The process of determining target content in different architectures is described in detail below.

[0139] In the architecture where the server cooperates with the terminal to execute the content generation method, the terminal runs a content generation service, and by inputting the first prompt information to the terminal, the target content corresponding to the first prompt information is generated in the content generation service, so that the terminal can display the target content, for example, the display interface of the terminal can display the target content, or the speaker of the terminal can play the target content. In this architecture, the target content can be generated in a variety of ways, and the various ways of generating the target content are described in detail below.

[0140] In the first determination method, the target content can be determined by the server. Specifically, after the terminal obtains the first prompt information, the terminal can send the first prompt information to the server so that the server can generate the target content corresponding to the first prompt information. Then the terminal can receive the target content sent by the server, that is, the server is responsible for executing all computing tasks to determine the target content, and the terminal is only responsible for sending and receiving data.

[0141] In the second determination method, the target content can be jointly determined by the terminal and the server. The terminal can perform preliminary processing on the first prompt information and upload the preliminary processing results to the server so that the server can determine the target content corresponding to the first prompt information. Then the terminal can receive the target content sent by the server, that is, the terminal is responsible for executing part of the computing task of determining the target content, and the server is responsible for executing another part of the computing task of determining the target content.

[0142] In the architecture where the terminal executes the content generation method, the terminal will run the content generation service, input the first prompt information into the terminal, and then generate the target content corresponding to the first prompt information in the content generation service, which is equivalent to the target content being determined by the terminal. The terminal can locally determine the target content corresponding to the first prompt information, that is, the terminal is responsible for executing all computing tasks to determine the target content.

[0143] In one possible implementation, the first prompt information may be determined based on the user's input content, for example, using the text input by the user through the keyboard as the first prompt information, or using the text selected by the user in the text template as the first prompt information, or converting the voice input by the user through a microphone into text, and then using the text as the first prompt information, or using the image, audio or video uploaded by the user through the conversation page as the first prompt information, wherein the image, audio or video may be generated in real time or in advance, or using the image, audio or video uploaded by the user through the conversation page as part of the first prompt information, and using the description text input by the user through the conversation page as another part of the first prompt information; in addition to the above methods, the first prompt information may also be determined in other ways, which are not limited in the embodiments of the present disclosure.

[0144] In one possible implementation, the target account and other user accounts are respectively associated with their own corresponding knowledge bases, which are used to store the private knowledge of the corresponding accounts so that the knowledge of different accounts is isolated from each other. The number of knowledge bases associated with an account can be one. For example, all the knowledge of an account can be stored in the same knowledge base, which can facilitate the unified management of knowledge.

[0145] In another possible implementation, an account may be associated with multiple knowledge bases. Therefore, based on the knowledge base pre-created by the target account, specifically one or more knowledge bases pre-created by the target account, for example, different categories of knowledge of an account may be stored in different knowledge bases. The category of knowledge may be determined by factors such as the field in which the knowledge is located or the source of the knowledge. By selecting knowledge in a specific field or knowledge from a specific source for content generation, the accuracy of the target content can be improved.

[0146] It is worth noting that multiple knowledge bases can be stored in the same storage system to facilitate management and reduce costs, and multiple knowledge bases can also be stored in different storage systems to enhance data isolation. Optionally, the knowledge base can be stored in a terminal or a server, or in both a terminal and a server, which is not limited in the embodiments of the present disclosure.

[0147] In one possible implementation, the content generation service may generate the target content through a variety of technical means. For example, in the content generation service, the first prompt information and the knowledge base may be input into a large language model for content generation to obtain the target content corresponding to the first prompt information, or relevant information may be retrieved and organized from the knowledge base through information retrieval to generate the target content corresponding to the first prompt information, or the target content corresponding to the first prompt information may be generated through predefined rules, conditions or templates; in addition, the target content may also be generated by other means, which are not limited in the embodiments of the present disclosure.

[0148] Specifically, the content generation method can be selected in a variety of ways, and the various selection methods are described in detail below.

[0149] In the first selection method, the selection is made through the selection result indicated by the selection information. Specifically, the selection information needs to be obtained first. The selection information may be information different from the first prompt information. The selection information may be in the form of one or more of text, image, audio and video. Then, by performing content recognition on the selection information, the selection result contained in the selection information can be determined, and the content generation method is automatically selected according to the selection result. For example, when the selection result indicates that content generation is performed based on a knowledge base, content generation is performed based on a knowledge base in the content generation service.

[0150] Exemplarily, the user can input first prompt information in voice form through the microphone, and then the speaker can play a selection prompt prompting the user to input selection information. The user can then input selection information in voice form through the microphone, and then perform content recognition on the input first prompt information to determine the selection result contained in the selection information. When the selection result indicates that content generation is to be performed in a knowledge base-based manner, then the knowledge base-based manner is selected in the content generation service to generate content. The entire process does not require the display of the page. Therefore, the user does not need to view the page or interact with the controls on the page, which can improve the interaction efficiency.

[0151] In addition, the selection information may also be a part of the first prompt information. Therefore, obtaining the first prompt information is equivalent to obtaining the selection information. By performing content recognition on the first prompt information, the selection result contained in the selection information can be determined. When the selection result indicates that content generation is to be performed based on a knowledge base, the knowledge base-based method is selected in the content generation service to generate content.

[0152] In the second selection method, the selection is made through preset preferences. Specifically, the user needs to set his or her preferences in advance, and then the content generation method is automatically selected based on the preferences. For example, when the preference is to generate content based on a knowledge base, the knowledge base-based method is selected in the content generation service.

[0153] In the third selection method, the selection is made through the classification results of the context. Specifically, the context needs to be obtained first, and then the classification is performed based on the context to obtain the classification results. The content generation method is automatically selected according to the classification results. For example, when the classification results indicate that the content is generated based on the knowledge base, the content generation service selects the knowledge base-based method to generate content, which realizes the senseless selection in the background and can improve the user experience. Among them, the context can be historical operations, historical conversation content, etc. The context can also be the current conversation content, that is, the context can also be part of the first prompt information. Therefore, obtaining the first prompt information is equivalent to obtaining the context.

[0154] In the fourth selection method, the selection can be made through the generation method selection control displayed on the session page. The following is a detailed description of the process of making a selection through the generation method selection control.

[0155] In one possible implementation, the first prompt information is input through a conversation page of a content generation service, and the conversation page displays a generation method selection control. When a knowledge base-based method is selected for content generation in the content generation service, target content corresponding to the first prompt information is generated in the content generation service based on a knowledge base pre-created by a target account. Specifically, when a knowledge base-based method is selected for content generation through the generation method selection control, target content corresponding to the first prompt information is generated in the conversation page based on a knowledge base pre-created by a target account.

[0156] Among them, the conversation page is an interface for human-computer interaction. The conversation page can be displayed on the display device of the terminal. The user can enter instructions and prompt information on the conversation page, and then the response results of the user input can be displayed on the conversation page. The conversation page can contain interactive elements and non-interactive elements. Interactive elements may include buttons, input boxes, drop-down lists, etc., and non-interactive elements may include static text, static images, labels, etc.

[0157] Optionally, the session page may include an interactive element for inputting the first prompt information. For example, the session page may include an input box for inputting the first prompt information, or the session page may include a file upload control for uploading a prompt file containing the first prompt information. For example, the file upload control may be a single-file input box, a multi-file input box, or a drag-and-drop upload control. Taking the drag-and-drop upload control as an example, the user may drag the prompt file directly from the desktop or folder to a designated area on the session page. In addition, the first prompt information may also be input into the session page in other ways, which is not limited in the embodiments of the present disclosure.

[0158] Among them, the generation method selection control is an interactive element, and the generation method selection control is used to provide the user with the function of selecting the generation method when generating content in the session page. Therefore, the user can choose to generate content based on the knowledge base through the generation method selection control, or choose other generation methods; the appearance of the generation method selection control can be arbitrary, for example, the outline of the generation method selection control can be a rectangle, a circle, etc., the outline of the generation method selection control can also be a specific icon, and the background of the generation method selection control can be a custom image, which is not limited in the embodiments of the present disclosure.

[0159] It should be noted that the user can first enter the first prompt information in the session page, and then select the knowledge base-based method to generate content through the generation method selection control in the session page; or, the user can first select the knowledge base-based method to generate content through the generation method selection control in the session page, and then enter the first prompt information in the session page; the present disclosed embodiment does not limit the user's operation process.

[0160] Based on this, by setting up a session page for the content generation service, the generation method selection control can be intuitively displayed in the session page, which is helpful for users to interact and make decisions. The target content can also be intuitively displayed in the session page, which helps users to quickly and accurately know the target content and improve the user experience. In addition, the interactive elements for entering the first prompt information can also be intuitively displayed in the session page, which helps to quickly and accurately enter the first prompt information.

[0161] Specifically, target content can be generated in a variety of scenarios, such as question-and-answer scenarios, writing scenarios, dialogue scenarios, translation scenarios, etc. Moreover, the form of the conversation page may be different in different scenarios, and the conversation page may also be called out in a variety of ways, which are not limited in the embodiments of the present disclosure. The following describes the form of the conversation page by taking some scenarios as examples.

[0162] For example, reference Figure 3 and Figure 4 , Figure 3 This is an optional operation diagram of a conversation page when asking questions in the first question-and-answer scenario provided by an embodiment of the present disclosure. Figure 4 An optional operation diagram of a conversation page when answering in the first question-and-answer scenario provided in an embodiment of the present disclosure.

[0163] In the first question-and-answer scenario, the user can ask questions in the text input box 311 of the knowledge base page 300 of the knowledge base, thereby displaying the conversation page 310, wherein the knowledge base page 300 is used to display reference content in the knowledge base, and the knowledge base page 300 can be an independent page.

[0164] When asking a question, the conversation page 310 displays a first generation mode selection control 312 and a second generation mode selection control 313. The user can enter the question text in the text input box 311. The question text is equivalent to the first prompt information. For example, the question text can be "What is the development trend of neural networks, and how is the effect of NLP?" The first generation mode selection control 312 and the second generation mode selection control 313 are both generation mode selection controls. The first generation mode selection control 312 is used to select a method based on a knowledge base, and the second generation mode selection control 313 is used to select a method based on a network search. Corresponding prompt texts are displayed on the first generation mode selection control 312 and the second generation mode selection control 313. For example, the prompt text of the second generation mode selection control 313 is "Full network search", and the prompt text of the first generation mode selection control 312 can be generated based on the question text. For example, the prompt text of the first generation mode selection control 312 can be "Answer based on the knowledge base: What is the development trend of neural networks, and how is the effect of NLP?"

[0165] When asking a question, when the user triggers the first generation mode selection control 312, the knowledge base-based method is selected for content generation. Then, the conversation page 310 will display the first reply text 314 generated by the content generation service. The first reply text 314 is equivalent to the target content. For example, the reply text can be "Based on these data, ..." The conversation page 310 will also display a text input box 311 when answering, so that the user can enter a new question text. When asking a question, when the user triggers the second generation mode selection control 313, the network search-based method is selected for content generation.

[0166] In addition, a continue generation control 315 may be displayed in the conversation page 310 . By triggering the continue generation control 315 , the next round of content generation may be performed based on the target content generated in the previous round and the question text currently input in the text input box 311 .

[0167] It is worth noting that after the conversation page 310 is displayed, the conversation page 310 can be closed. For example, the conversation page 310 is located on the upper layer of the knowledge base page 300 and covers a local area of ​​the knowledge base page 300. The user can close the conversation page 310 by clicking on the uncovered area of ​​the knowledge base page 300.

[0168] For example, reference Figures 5 to 7 , Figure 5 An optional operation diagram of an initial conversation page in a writing scenario provided by an embodiment of the present disclosure is shown in FIG. Figure 6 This is an optional operation diagram of a conversation page when inputting writing requirements in a writing scenario provided by an embodiment of the present disclosure. Figure 7An optional operation diagram of a conversation page for a preliminary reply in a writing scenario provided by an embodiment of the present disclosure.

[0169] In the writing scenario, the conversation page 310 can be an independent page and does not need to be called out based on the knowledge base page. The conversation page 310 displays a writing entry control 501 for accessing the question interface. The prompt text below the writing entry control 501 is "Smart Writing". When the user triggers the writing entry control 501, a text input box 511, a submission control 512, a menu control 513, a third generation method selection control 514 and a fourth generation method selection control 515 are displayed. The user can enter the writing requirements in the text input box 511. The writing requirements are equivalent to the first prompt information. For example, the writing requirements can be "Help me write an opening speech for the event, about 500 words", and the menu control 513 is used to display the third generation method selection control 514 and the fourth generation method selection control 515. The third generation mode selection control 514 and the fourth generation mode selection control 515 are hidden before the menu control 513 is triggered. When the user triggers the menu control 513, the third generation mode selection control 514 and the fourth generation mode selection control 515 will be displayed. The third generation mode selection control 514 and the fourth generation mode selection control 515 are both generation mode selection controls. The third generation mode selection control 514 is used to select a method based on a temporary file, and the fourth generation mode selection control 515 is used to select a method based on a knowledge base.

[0170] When asking a question, when the user triggers the fourth generation method selection control 515, the method based on the knowledge base is selected. At this time, there is no need to specify the reference content, and the generation is directly based on all the reference content of the knowledge base by default. Alternatively, a page for selecting specified reference content in the knowledge base can be displayed. The submission control 512 is used to submit writing requirements. When the user triggers the submission control 512, the writing requirements are submitted and content generation is performed based on the selected method. Then, when answering, the conversation page 310 will display the second reply text 516 generated by the content generation service. The second reply text 516 is equivalent to the target content. When asking a question, when the user triggers the third generation method selection control 514, the method based on temporary files is selected for content generation. At this time, there is no need to specify the temporary file, and the generation is directly based on the pre-configured temporary file by default. Alternatively, a page for selecting a specified temporary file can be displayed.

[0171] In addition, when asking a question, the conversation page 310 also displays a first template selection control 502, a second template selection control 503, a third template selection control 504 and a fourth template selection control 505. The user can select a corresponding writing template by triggering one of the first template selection control 502, the second template selection control 503, the third template selection control 504 and the fourth template selection control 505. For example, the writing template of the first template selection control 502 is not limited, the writing template of the second template selection control 503 is a paper, the writing template of the third template selection control 504 is an essay, and the writing template of the fourth template selection control 505 is an article. These writing templates can all be pre-configured. In addition to the above-mentioned writing templates, other writing templates can also be pre-configured, and the embodiments of the present disclosure are not limited here.

[0172] In one possible implementation, since there can be multiple ways to generate the target content, and content generation based on a knowledge base is only one of the multiple generation methods, the generation method can also include a network search-based method or a temporary file-based method, etc. Therefore, the generation method selection control can be associated with the generation method, that is, the generation method selection control can indicate the corresponding generation method.

[0173] For example, a session page displays multiple generation method selection controls, one of which is used to select a knowledge base-based method, and another is used to select other methods. At this time, when the generation method selection control used to select the knowledge base-based method is triggered, it is considered that the knowledge base-based method is selected for content generation. In addition, multiple generation methods can be selected at the same time, such as selecting a knowledge base-based method and a temporary file-based method for content generation. During the generation process, content can be generated based on reference knowledge in the knowledge base and actual data, information or context in the temporary file. Compared with selecting only one generation method, the generation quality of the target content can be further improved.

[0174] For another example, the session page may also only display a generation method selection control for selecting a knowledge base-based method. When the generation method selection control is triggered, it is considered that the knowledge base-based method is selected for content generation, and when the generation method selection control is not triggered, it is considered that the default generation method is selected for content generation, where the default generation method can be a method based on network search.

[0175] Specifically, the session page can also usually display the prompt text of the generation method selection control, and the generation method selection control can be in various forms. For example, the generation method selection control can be in the form of a button, check box, slider or other controls for selecting a given option. For another example, the generation method selection control can also be in the form of a file selection control or a list selection control or other controls for specifying knowledge content.

[0176] The following describes the form of the generation method selection control by taking some forms as an example.

[0177] First, taking the generation method selection control as a check box as an example, when the check box for selecting the knowledge base-based method is selected, it is considered that the knowledge base-based method is selected for content generation, and when the check box is not selected, it is considered that the knowledge base-based method is not selected for content generation. The prompt text of the check box can be set to "Generate content based on the knowledge base when the check box is selected".

[0178] Then, take the button as an example and refer to Figure 3 , the conversation page 310 displays a first generation method selection control 312 and a second generation method selection control 313. The first generation method selection control 312 and the second generation method selection control 313 are both generation method selection controls. The first generation method selection control 312 is used to select a knowledge base-based method, and the second generation method selection control 313 is used to select a network search-based method. Therefore, when the first generation method selection control 312 is triggered, it is considered that the knowledge base-based method is selected for content generation, and when the second generation method selection control 313 is triggered, it is considered that the network search-based method is selected for content generation.

[0179] Then, refer to Figure 6, the session page 310 displays the third generation mode selection control 514 and the fourth generation mode selection control 515 in the form of a selection menu. The third generation mode selection control 514 and the fourth generation mode selection control 515 are both generation mode selection controls. The third generation mode selection control 514 is used to select a mode based on a temporary file, and the fourth generation mode selection control 515 is used to select a mode based on a knowledge base. Therefore, when the third generation mode selection control 514 is triggered, a mode based on a temporary file is selected for content generation. In this case, there is no need to specify a temporary file, and the content is generated based on a pre-configured temporary file by default. Alternatively, a page for selecting a specified temporary file may be displayed; and when the fourth generation mode selection control 515 is triggered, a mode based on a knowledge base is selected for content generation. In this case, there is no need to specify a reference content, and the content is generated based on all reference contents of the knowledge base by default. Alternatively, a page for selecting specified reference content in the knowledge base may be displayed.

[0180] In one possible implementation, there are multiple generation method selection controls for selecting a knowledge base-based method. Since multiple knowledge bases can be stored in different storage systems, the generation method selection controls can be associated with the storage systems so that the session page can display the generation method selection controls corresponding to each storage system. When any generation method selection control for selecting a knowledge base-based method is triggered, it is considered that the knowledge base-based method is selected for content generation.

[0181] In a possible implementation, the generation method selection control may be located in the first-level menu of the session page, that is, the generation method selection control will be directly displayed, and the user can directly interact with the generation method selection control without additional steps, which can improve efficiency. In addition, the generation method selection control may also be located in the multi-level menu of the session page, and the user needs to expand the multi-level menu before the generation method selection control is displayed, so as to avoid the session page from being complicated due to displaying too many controls, and keep the session page concise and clear.

[0182] For example, again refer to Figure 6 The menu control 513 in the conversation page 310 is a secondary menu. The third generation method selection control 514 and the fourth generation method selection control 515 will be displayed only when the menu control 513 is expanded. When the secondary menu is not expanded, these two controls will not be displayed, so that the conversation page 310 remains concise and clear.

[0183] In a possible implementation manner, the content generation method further includes: displaying at least one first reference link in the session page.

[0184] Among them, the first reference link is used to access the reference content used to generate the target content in the knowledge base. The first reference link is specifically an interactive element or a non-interactive element. When the first reference link is an interactive element, the first reference link can be a hyperlink control. The session page allows the user to quickly jump to access the reference content by clicking the first reference link, that is, to access the reference content by clicking the first reference link. When the first reference link is a non-interactive element, the user is not allowed to click the first reference link, and needs to access the reference content through other methods. For example, the user enters the first reference link into the path input bar to directly access the reference content.

[0185] The first reference link may be embedded in one or more of the elements such as text, image, audio, and video, which is not limited in the embodiment of the present disclosure.

[0186] Based on this, since the amount of knowledge in the knowledge base is usually large, and only part of the knowledge is helpful for the generation of target content, when generating content, usually only part of the knowledge in the knowledge base is referenced to generate the target content. The knowledge referenced in the session content generation process can be determined as reference content. The reference content can be selected by the user or automatically selected. Therefore, the reference content can be regarded as the basis for generating the target content, that is, the reference content is used to generate the target content. By displaying the first reference link in the session page, it is convenient for users to quickly access the reference content through the first reference link, so that users can know the basis for generating the target content, which helps users analyze the credibility and accuracy of the target content.

[0187] It is worth noting that when the number of reference contents referred to by the session content generation process is one, the first reference link of the reference content is displayed in the session page, that is, the number of first reference links displayed is one, and when the number of reference contents referred to by the session content generation process is multiple, the first reference links of each reference content are displayed in the session page, that is, the number of first reference links displayed is multiple.

[0188] For example, refer again to Figure 3 and Figure 4 The session page 310 displays a plurality of first reference links 316, each of which is embedded with a schematic image and a schematic text. By triggering any first reference link 316, the user can quickly jump to access the reference content, for example, to jump to access a specific web page, document, audio, or video content.

[0189] In a possible implementation, the content generation method further includes: when content generation based on network search is selected in the content generation service, generating target content corresponding to the first prompt information in the content generation service based on search results of the network search.

[0190] Similar to the knowledge base-based method, the conversation page may display a generation method selection control for selecting a network search-based method. When content generation based on network search is selected in the content generation service, target content corresponding to the first prompt information is generated in the content generation service based on the search results of the network search. Specifically, when content generation based on network search is selected through the generation method selection control, target content corresponding to the first prompt information is generated in the conversation page based on the search results of the network search.

[0191] Among them, choosing a method based on network search for content generation specifically means that the generation process of the target content needs to refer to the search results of the network, and by obtaining real-time, extensive, and multi-dimensional search results, it is ensured that the generation process of the target content can be more accurate, comprehensive, and keep up with the latest developments; in addition, the network search can be conducted based on the first prompt information, so that the search results are related to the first prompt information, ensuring that the generated target content not only has reference value, but also can respond to the core needs in the first prompt information.

[0192] In a possible implementation, the content generation method further includes: when content generation based on a temporary file is selected in the content generation service, generating target content corresponding to the first prompt information in the content generation service based on the temporary file.

[0193] Similar to the knowledge base-based method, the session page may display a generation method selection control for selecting a temporary file-based method. When content generation based on temporary files is selected in the content generation service, the target content corresponding to the first prompt information is generated in the content generation service based on the temporary files. Specifically, when content generation based on temporary files is selected through the generation method selection control, the target content corresponding to the first prompt information is generated in the session page based on the temporary files.

[0194] Among them, choosing to generate content based on a temporary file specifically means that the generation process of the target content needs to refer to the information of the temporary file, so that the generated target content can ensure a high degree of match with the actual data, information or context in the temporary file, thereby effectively improving the quality of the target content; in addition, the information referenced in the temporary file can be information related to the first prompt information, thereby ensuring that the generated target content not only has reference value, but also can respond to the core requirements in the first prompt information.

[0195] It should be noted that temporary files refer to files temporarily used in the process of generating target content. After the target content is generated, temporary files are usually deleted and will not be retained for a long time, which can avoid data leakage and improve security. Temporary files can be any type of file, for example, temporary files can be one or more of text files, image files, audio files and video files. It is worth noting that temporary files can be located in the terminal or server, which is not limited in the embodiments of the present disclosure.

[0196] In one possible implementation, when content generation based on temporary files is selected in the content generation service, target content corresponding to the first prompt information is generated in the content generation service based on the temporary files. Specifically, when content generation based on temporary files is selected in the content generation service, a file selection page is displayed, wherein the file selection page is used to display pre-stored temporary files; in response to the operation of selecting at least one temporary file in the file selection page, the target content corresponding to the first prompt information is generated in the session page based on the selected temporary file.

[0197] The generation mode selection control for selecting the temporary file-based mode may be a control for specifying knowledge content. For example, the generation mode selection control may be in the form of a file selection control or a list selection control. Specifically, when the generation mode selection control for selecting the temporary file-based mode is triggered, a file selection page is displayed, and it is considered that the temporary file-based mode is selected for content generation.

[0198] For example, again refer to Figure 6 The session page 310 displays a third generation method selection control 514, which is used to select a method based on temporary files. When the third generation method selection control 514 is triggered, it is considered that the temporary file-based method is selected for content generation, and then a page for selecting a specified temporary file can be displayed.

[0199] Specifically, refer to Figures 9 to 11 , Fig. 9 This is an optional operation diagram of a conversation page when asking questions in the second question-and-answer scenario provided by an embodiment of the present disclosure. Fig.10 This is an optional operation diagram of a conversation page for the first reply in the second question-and-answer scenario provided by an embodiment of the present disclosure. Fig.11 An optional operation diagram of a conversation page when replying again in the second question-and-answer scenario provided in an embodiment of the present disclosure.

[0200] Among them, in the second question-and-answer scenario, the user can first ask a question on the conversation page 310, and then view the answer on the conversation page 310. When asking a question, the conversation page 310 displays a first text input box 901, and the user can enter the question text in the first text input box 901. The question text is equivalent to the first prompt information. For example, the question text can be "What is the development trend of neural networks, and what is the effect of NLP". After the user enters the question text in the first text input box 901, content generation will be performed by the default generation method.

[0201] In addition, the conversation page 310 may also display a file upload control 902 and a screenshot control 903. The user may upload a file for interpretation by triggering the file upload control 902, and the user may start the screenshot function by triggering the screenshot control 903. The captured image may also be used as the first prompt information.

[0202] Taking the default generation method based on network search as an example, when replying for the first time, the conversation page 310 will display the question content 904, the third reply text 905, the second text input box 906 and the fifth generation method selection control 907. The question content 904 is the same as the question text entered in the first text input box 901. The third reply text 905 is equivalent to the target content. The user can enter a new question text in the second text input box 906. The fifth generation method selection control 907 is a generation method selection control. The fifth generation method selection control 907 is used to select a knowledge base-based method. Therefore, when the user triggers the fifth generation method selection control 907, the knowledge base-based method is selected to generate content again, and the newly generated reply is displayed in the conversation page 310.

[0203] When replying again, the conversation page 310 will display the fourth reply text 911, the question content 904, the second text input box 906 and the sixth generation method selection control 910. The fourth reply text 911 is generated based on the knowledge base method, and the sixth generation method selection control 910 is a generation method selection control. The sixth generation method selection control 910 can be used to select a method based on network search. Therefore, when the user triggers the sixth generation method selection control 910, the method based on network search is selected to generate content again.

[0204] In addition, after replying based on the network search, the conversation page 310 displays multiple second reference links 909, and the search results of the network search can be quickly jumped to by triggering any second reference link 909. After replying based on the knowledge base, the conversation page 310 can also display multiple first reference links 316, and the reference content can be quickly jumped to by triggering any first reference link 316.

[0205] In one possible implementation, in response to an operation of selecting at least one first reference link in a conversation page, the selected first reference link is marked in the conversation page; in response to an operation of regenerating target content in the conversation page, the target content is regenerated in the conversation page based on the reference content corresponding to the selected first reference link.

[0206] Among them, after displaying at least one first reference link in the conversation page, the user can select the first reference link, mark the selected first reference link in the conversation page, or select the selected first reference link, or display the selected first reference link in bold, or change the color of the selected first reference link, etc. For example, the current target content is generated based on three reference contents in the knowledge base, and the first reference links corresponding to the three reference contents will be displayed in the conversation page. Exemplarily, the user can select two of the first reference links, and then regenerate the target content in the conversation page based on the reference contents corresponding to the two selected first reference links. In this way, based on the target content generated by the knowledge base, a more refined result can be further generated according to the demand. Moreover, since the first reference link of the reference content used to generate the current target content is displayed in the conversation page, the user only needs to directly select the first reference link in the conversation page to determine the reference content actually needed and regenerate the target content according to the demand, thereby quickly adjusting the generated target content.

[0207] It is understandable that each time a target content is generated, a first reference link corresponding to the reference content used to generate the current target content will be displayed in the session page. Therefore, the user can generate the target content in multiple rounds by selecting the first reference link until the generated target content meets his or her own requirements.

[0208] In one possible implementation, in response to an operation of regenerating target content in a session page, specifically, in response to an operation of triggering a regeneration control, a regeneration control is displayed in the session page, and the regeneration control is used to regenerate new target content. When the user triggers the regeneration control, the target content is regenerated in the session page.

[0209] In one possible implementation, when a knowledge base-based method is selected for content generation through a generation method selection control, target content corresponding to the first prompt information is generated in a session page based on a knowledge base pre-created by a target account. Specifically, when a knowledge base-based method is selected for content generation through a generation method selection control, a knowledge base page is displayed, wherein the knowledge base page is used to display reference content in a knowledge base pre-created by a target account; in response to an operation of selecting at least one reference content in the knowledge base page, target content corresponding to the first prompt information is generated in the session page based on the selected reference content.

[0210] The generation mode selection control for selecting the knowledge base-based mode may be a control for specifying knowledge content. For example, the generation mode selection control may be in the form of a file selection control or a list selection control. Specifically, when the generation mode selection control for selecting the knowledge base-based mode is triggered, the knowledge base page is displayed, and it is considered that the knowledge base-based mode is selected for content generation.

[0211] Among them, the knowledge base page is an interface for human-computer interaction. The knowledge base page can be located on the upper layer of the session page, or the session page can be hidden and only the knowledge base page can be displayed to ensure that the content of the knowledge base page can be displayed completely. The user can select at least one reference content in the knowledge base page.

[0212] The operation of selecting the reference content may specifically be triggering a control or directly operating the content, etc. Other operations may also be used, which are not limited in the embodiments of the present disclosure.

[0213] With respect to triggering a control, a first reference control corresponding to the reference content is displayed in the knowledge base page, and the user's operation of triggering the first reference control in the session page can be regarded as an operation of selecting the reference content. Therefore, when the user triggers the first reference control, the reference content corresponding to the first reference control is selected, wherein triggering the first reference control may be clicking, inputting, dragging, hovering, etc., on the first reference control.

[0214] For example, the first reference control is a first reference button, and the first reference button corresponding to the reference content is displayed on the knowledge base page, and the reference content is selected by clicking the first reference button; for another example, the first reference control is a first input box, and the name of the reference content and the first input box are displayed on the knowledge base page, and the reference content is selected by entering the name of the reference content in the first input box; for another example, a selected area and a candidate area are displayed on the knowledge base page, and a draggable first reference control corresponding to each reference content is placed in the candidate area, and the reference content is selected by dragging the first reference control from the candidate area to the selected area; for another example, the first reference control corresponding to each reference content is displayed on the knowledge base page, and the reference content is selected by hovering the cursor over the first reference control.

[0215] In addition, the first reference control may also be a third reference link, and the reference content may be selected by performing corresponding operations on the third reference link. For example, single-clicking the third reference link selects the reference content corresponding to the third reference link, while double-clicking the third reference link opens the third reference link to preview the reference content.

[0216] For direct operations on content, users' operations such as selecting all, partially selecting, single-clicking, double-clicking, bolding or highlighting the target content are all direct operations on the content. Usually, it is necessary to set operation rules to specify which operations can be used as operations for selecting reference content. For example, after specifying the operation of highlighting reference content as the operation for selecting reference content through operation rules, when the user highlights the reference content, the reference content is selected.

[0217] It should be noted that the types of reference content can be various, for example, the reference content can be text, picture, video, voice, etc.

[0218] Based on this, by displaying the reference content in the knowledge base pre-created by the target account on the knowledge base page, all selectable reference content can be displayed intuitively, and the user can quickly and independently select at least one reference content on the knowledge base page. After the user selects the reference content, the target content corresponding to the first prompt information can be generated in the session page based on the selected reference content. In this way, the reference content can be used as the basis for generating the target content, so that the target content can include the basic facts, rules and other content contained in the reference content.

[0219] For example, again refer to Figure 6 , and refer to Fig.12 , Fig.12A schematic diagram of an optional operation of a knowledge base page in a second conversation scenario provided by an embodiment of the present disclosure. When a user triggers the fourth generation mode selection control 515 on the conversation page 310, the knowledge base page 300 is displayed, and the knowledge base page 300 displays a plurality of third reference links 1211. The user can select reference content by triggering the third reference link 1211 on the knowledge base page 300, for example, by clicking the third reference link 1211 to select the corresponding reference content.

[0220] In one possible implementation, when a knowledge base-based method is selected for content generation, a knowledge base page may be displayed first, and a user may select reference content on the knowledge base page. In response to the operation of selecting at least one reference content on the knowledge base page, a session page may be displayed, and then the user may enter first prompt information on the session page. In response to the operation of generating target content, target content corresponding to the first prompt information is generated on the session page based on the selected reference content.

[0221] In one possible implementation, the knowledge base pre-created by the target account contains multiple reference contents, and the knowledge base page is displayed. Specifically, the correlation score between each reference content and the first prompt information is determined, and then the reference contents are sorted in descending order based on the correlation score. Then, the knowledge base page is displayed based on the sorted reference contents, so that the knowledge base page can display the sorted reference contents, which is convenient for users to quickly locate reference contents with higher correlation and select reliable reference contents, thereby improving the accuracy of the target content; optionally, the knowledge base page will also display the correlation score of each reference content, which further helps users select reliable reference contents.

[0222] In a possible implementation, the content generation method further includes: in response to an operation of extending the target content in the session page, performing content expansion on the target content in the content generation service to obtain the extended content.

[0223] The operation of expanding the target content may specifically be triggering a control or directly operating the content, etc. Other operations may also be used, which are not limited in the embodiments of the present disclosure.

[0224] For triggering the control, an extension generation control is displayed in the session page, and the user's operation of triggering the extension generation control in the session page can be used as an operation of expanding the target content in the session page. Therefore, when the user triggers the extension generation control, the target content is expanded, wherein triggering the extension generation control may be clicking, inputting, or dragging the extension generation control.

[0225] For direct operations on content, users can select all, select part, click, double-click, bold or highlight the target content. These operations are all direct operations on the content. Usually, it is necessary to set operation rules to specify which operations can be used as operations to expand the target content in the session page. For example, after specifying the operation of highlighting the target content as the operation of expanding the target content in the session page through operation rules, when the user highlights the target content, the target content is expanded.

[0226] Based on this, the target content generated first is used as the preview content of the final generation result, that is, the target content is usually brief and general. For example, the target content can be the outline of an article, the abstract of an article, the outline of an audio, the outline of a video, etc. When the user thinks the target content is appropriate, the user can extend the target content in the session page, thereby extending the target content in the content generation service to obtain the extended content, that is, the final generation result. For example, the extended content can be a complete article, audio or video, etc.

[0227] For example, reference Figure 7 and Figure 8 , Figure 8 An optional operation diagram of a conversation page for further replying in a writing scenario provided in an embodiment of the present disclosure, in which, in a preliminary reply, the conversation page 310 displays an extension generation control 517, a regeneration control 518 and a discard control 519, and when the extension generation control 517 is triggered, the target content is expanded in the content generation service, and the extended reply text 520 is displayed in the conversation page 310, and the reply text 520 is equivalent to the extended content; when the regeneration control 518 is triggered, a new reply is regenerated in the conversation page 310; when the discard control 519 is triggered, the second reply text 516 generated in the preliminary reply is discarded.

[0228] In a possible implementation, before acquiring the first prompt information input to the content generation service, the content generation method further includes: in response to an operation of adding content to the knowledge base, adding reference content browsed or processed by the target account to the knowledge base.

[0229] The operation of adding content to the knowledge base may specifically be triggering a control or directly operating the content, etc. Other operations may also be used, which are not limited in the embodiments of the present disclosure.

[0230] With respect to triggering the control, a second reference control corresponding to the reference content is displayed in the session page, and the user's operation of triggering the second reference control in the session page can be regarded as an operation of adding content to the knowledge base. Therefore, when the user triggers the second reference control, the reference content browsed or processed by the target account is added to the knowledge base, wherein triggering the second reference control may be clicking, inputting, dragging, etc., on the second reference control.

[0231] For example, the second reference control is a second reference button, and a second reference button corresponding to the reference content is displayed on the knowledge base page, and the reference content is added to the knowledge base by clicking the second reference button; for another example, the second reference control is a second input box, and a second input box is displayed on the knowledge base page, and the reference content is added to the knowledge base by entering specified text in the second input box, and the specified text can be set to "Add"; for another example, the second reference control is a drag upload control, and the drag upload control is displayed on the knowledge base page, and the reference content is added to the knowledge base by directly dragging the reference content to the specified area of ​​the drag upload control, or the content import page is displayed after the drag upload control is triggered, and then the reference content is directly dragged to the drag content import page to add the reference content to the knowledge base.

[0232] For direct operations on content, users can directly operate on reference content by selecting all, partially selecting, single-clicking, double-clicking, bolding or highlighting the target content. Usually, it is necessary to set operation rules to specify which operations can be used as operations for adding content to the knowledge base. For example, after specifying the operation of highlighting reference content as an operation for adding content to the knowledge base through operation rules, when the user highlights the reference content, the reference content browsed or processed by the target account is added to the knowledge base.

[0233] Based on this, since the reference content is the content browsed or processed by the target account, the reference content is the private data of the target account, and the reference content will contain the private professional knowledge of the target account. Therefore, after adding the reference content to the knowledge base, when generating content based on the knowledge base, it is possible to generate more specialized and more in line with the target account's private knowledge field, that is, to generate more focused and professional target content; on this basis, the types and sources of content browsed or processed by the target account are very diverse, so the types and sources of reference content are also very diverse. For example, the reference content may include web pages, files, articles, etc. browsed by the target account. The reference content may also include files uploaded by the target account, search results, first prompt information inputted in history, or target content generated in history, etc. Adding reference content of diverse types and sources to the knowledge base can obtain a multi-source knowledge base, so that data of different types and sources can be stored and traced back. The multi-source knowledge base can provide comprehensive and reliable reference knowledge when generating content based on the knowledge base. Therefore, it can further improve the professionalism and accuracy of the target content, so that the target content can better meet user needs.

[0234] In one possible implementation, in response to an operation of adding content to a knowledge base, reference content browsed or processed by a target account is added to the knowledge base, which may specifically be a content import page of the knowledge base displayed; in response to an operation of adding content to the knowledge base in the content import page, reference content imported by the target account is added to the knowledge base.

[0235] Among them, the content import page is an interface for human-computer interaction. Users can add content to the knowledge base in the content import page. The operation of adding content to the knowledge base may include clicking, inputting or dragging, etc., which is not limited in the embodiments of the present disclosure.

[0236] Based on this, the reference content of the target account can be imported through the content import page first. The imported reference content is equivalent to the reference content processed by the target account. Then, the reference content imported by the target account can be added to the knowledge base through the content import page. Adding reference content by importing eliminates the need for users to manually enter reference content. Especially when uploading reference content in batches, a large amount of reference content can be imported at one time without having to enter it one by one, which can greatly reduce the time spent on adding reference content and avoid adding erroneous reference content due to human error, thereby ensuring the accuracy of the reference content.

[0237] It should be noted that the operation of adding content to the knowledge base in the content import page may be a single operation or a series of operations, which is not limited in the embodiments of the present disclosure.

[0238] For example, refer to Fig.13 , Fig.13 An optional operation diagram of the content import page provided in the embodiment of the present disclosure, the content import page 1310 displays an upload area 1311, and the user can drag and drop reference content from the desktop or folder directly to the upload area 1311 to import the reference content. Moreover, since the content import page 1310 can directly display the content in the knowledge base, the user only needs to drag the reference content to the upload area 1311, and the reference content will be automatically added to the knowledge base, and no additional operations can be performed. Therefore, the user's dragging operation is an operation to add content to the knowledge base in the content import page 1310, so that adding content to the knowledge base can be completed through a single operation, which can improve the efficiency of adding reference content.

[0239] For example, refer to Fig.14 , Fig.14 Another optional operation diagram of the content import page provided in the embodiment of the present disclosure is shown. The content import page 1410 can display multiple reference contents 1411 to be imported, an open control 1412, and a cancel control 1413. The user can first select the reference content 1411 to be imported in the content import page 1410, and then add the imported reference content to the knowledge base by triggering the open control 1412. Therefore, the operation of adding content to the knowledge base in the content import page 1410 is completed in two steps, and the addition of content to the knowledge base is completed through multiple operations, avoiding the addition of wrong reference content, and improving the accuracy of the reference content. In addition, when the user triggers the cancel control 1413, the currently displayed content import page 1410 will be closed, and the user can re-call the content import page 1410 by triggering the add button 1421 in the knowledge base page 300.

[0240] In one possible implementation, in response to an operation of adding content to a knowledge base, reference content browsed or processed by a target account is added to the knowledge base. Specifically, a session page of a content generation service may be displayed, and second prompt information entered in the session page may be obtained; when content generation is performed in the session page, reference content corresponding to the second prompt information is generated in the session page; in response to an operation of adding content to the knowledge base in the session page, reference content generated by the target account through the session is added to the knowledge base.

[0241] Among them, the second prompt information is used to guide the content generation service to generate content. Similar to the first prompt information, the second prompt information can also be in the form of one or more of text, image, audio and video. The second prompt information can also be in other forms, which is not limited in the embodiments of the present disclosure.

[0242] Based on this, the reference content generated by the session is first obtained through content generation. The reference content generated by the session is equivalent to the reference content processed by the target account. Then the reference content generated by the session is added to the knowledge base. Similar to the target content generation process, the reference content can also be generated based on the knowledge base, network search or temporary files. Combined with the limitation that the reference content is generated before the target content is generated, it can be seen that the reference content is equivalent to the historically generated target content. Similar to the description of the target content generation process, the method based on the knowledge base, network search or temporary files is selected to ensure that the generated reference content not only has reference value, but also can respond to the core needs in the second prompt information; after the reference content is generated in the session page, the user can perform the operation of adding content to the knowledge base in the session page, thereby effectively adding the reference content generated by the session to the knowledge base, so that the knowledge base can store knowledge generated from the session, can flexibly add knowledge and improve the diversity of knowledge sources.

[0243] Specifically, the operation of adding content to the knowledge base may be to trigger a control or to directly operate the content, etc. Other operations may also be used, which are not limited in the embodiments of the present disclosure.

[0244] Taking triggering a control as an example, you can refer to Figure 8 , the conversation page 310 displays a third adding control 521. When the user triggers the third adding control 521, the reply text 520 is added to the knowledge base. The reply text 520 may specifically be reference content generated through the conversation.

[0245] You can also refer to it again Fig.10 When replying for the first time, the conversation page 310 displays a fourth add control 908. When the user triggers the fourth add control 908, the third reply text 905 is added to the knowledge base. Fig.11 When replying again, the conversation page 310 also displays a fourth adding control 908. When the user triggers the fourth adding control 908, the fourth reply text 911 is added to the knowledge base.

[0246] Also see Fig.15 , Fig.15 An optional operation diagram of a media page provided for an embodiment of the present disclosure, wherein the first media page 1510 can be a note page, displaying note content 1511 and a fifth add control 1512, and the note content 1511 can be used as reference content. When the user triggers the fifth add control 1512, the note content 1511 is added to the knowledge base.

[0247] In addition, when Fig. 9 When generating content in the session page shown, you can jump to Fig.15The first media page 1510 shown in the figure displays the target content, so that the target content can be used to quickly create notes, thereby improving the efficiency of human-computer interaction.

[0248] Also see Fig.16 , Fig.16 Another optional operation diagram of the media page provided for the embodiment of the present disclosure, wherein the second media page 1610 can be a web page, displaying web page content 1611 and a sixth add control 1612, and the web page content 1611 can be used as reference content. When the user triggers the sixth add control 1612, the web page content 1611 is added to the knowledge base.

[0249] The operation of expanding the target content may specifically be triggering a control or directly operating the content, etc. Other operations may also be used, which are not limited in the embodiments of the present disclosure.

[0250] For triggering the control, an extension generation control is displayed in the session page, and the user's operation of triggering the extension generation control in the session page can be used as an operation of expanding the target content in the session page. Therefore, when the user triggers the extension generation control, the target content is expanded, wherein triggering the extension generation control may be clicking, inputting, or dragging the extension generation control.

[0251] For direct operations on content, users can select all, select part, click, double-click, bold or highlight the target content. These operations are all direct operations on the content. Usually, it is necessary to set operation rules to specify which operations can be used as operations to expand the target content in the session page. For example, after specifying the operation of highlighting the target content as the operation of expanding the target content in the session page through operation rules, when the user highlights the target content, the target content is expanded.

[0252] In one possible implementation, in response to an operation of adding content to a knowledge base, reference content browsed or processed by a target account is added to the knowledge base, which may specifically be a media page displayed for browsing by the target account; in response to an operation of adding content to the knowledge base in a media page, reference content in the media page is added to the knowledge base.

[0253] Among them, the media page is an interface for human-computer interaction, and the media page can display the content of media such as web pages and documents.

[0254] Based on this, the media page browsed by the target account is first displayed. The reference content in the media page is equivalent to the reference content browsed by the target account. Then the reference content in the media page is added to the knowledge base. Specifically, the user can add content to the knowledge base in the media page, thereby effectively adding the reference content of the media page to the knowledge base, so that the knowledge base can store knowledge derived from the media page, can flexibly add knowledge and improve the diversity of knowledge sources.

[0255] Specifically, users can browse media pages in a client running a content generation service. In this case, the client may be integrated with a browser kernel so that the client can access and browse web pages, or the client may be integrated with a document viewing kernel so that the client can view documents. Users can also browse media pages in other clients. In this case, the client used to browse media pages may be deployed with a plug-in for adding knowledge, and users can use the plug-in to add content to the knowledge base in the media page.

[0256] In one possible implementation, a media page browsed by the target account is displayed, and specifically, it may be a session page that displays a content generation service. When content generation is performed based on a network search in the session page, at least one second reference link is displayed after the content is generated in the session page; in response to an operation of opening the second reference link, the media page browsed by the target account is displayed.

[0257] Among them, the second reference link is used to access the search results of the network search. The second reference link is specifically an interactive element or a non-interactive element. When the second reference link is an interactive element, the second reference link can be a hyperlink control. The session page allows the user to quickly jump to access the reference content by clicking the second reference link, that is, to access the reference content by clicking the second reference link. When the second reference link is a non-interactive element, the user is not allowed to click the second reference link, and needs to access the reference content through other methods, for example, the user enters the second reference link into the address bar to directly access the reference content.

[0258] Based on this, when the user selects a network search-based method to generate content through the session page, the session page can generate corresponding session content. At the same time, the session page will also display a second reference link. This is because when generating content, the search results will be determined through a network search first, and then the corresponding session content will be generated based on the search results. Therefore, the search results can be used as the basis for generating the currently generated session content. By displaying the second reference link in the session page, the user can quickly access the search results indicated by the second reference link by opening the second reference link. The search result can be a media page. Since the media page will be displayed and browsed by the user, the media page can also be regarded as a media page browsed by the target account. Therefore, by displaying the second reference link, it is helpful for the user to quickly locate the search results referenced in the session content generation process, that is, to locate the relevant knowledge referenced in the session process. After the media page is displayed, the relevant knowledge can be quickly added to the knowledge base, thereby improving the efficiency and accuracy of knowledge addition.

[0259] It is worth noting that when the number of search results referenced in the process of generating session content is one, the second reference link of the search result is displayed in the session page, that is, the number of second reference links displayed is one, and when the number of search results referenced in the process of generating session content is multiple, the second reference link of each search result is displayed in the session page, that is, the number of second reference links displayed is multiple.

[0260] In one possible implementation, in addition to being all the original content in the media, the reference content may also be in response to an operation of adding content to the knowledge base, obtain the media browsed or processed by the target account, parse the media to obtain the reference content, and add the reference content to the knowledge base.

[0261] Specifically, the operation of adding content to the knowledge base can refer to the previous explanation and will not be repeated here. The media browsed or processed by the target account is a carrier that can carry data, such as a web page or a document. Parsing the media is to capture data from the media. This process can include operations such as formatting, cleaning, and extraction. Correspondingly, the reference content is valuable information in the media. Taking the media as a web page as an example, the reference content obtained by parsing the web page can be the title, text, and pictures of the web page. By parsing the media to obtain the reference content and then adding the reference content to the knowledge base, useless information can be removed, the data redundancy of the reference content can be reduced, and the knowledge carried by the reference content can be more accurate. When generating content based on the reference content, the generation efficiency and accuracy can be effectively improved.

[0262] In addition, after parsing the media to obtain reference content, the reference content can be input into the large language model for understanding analysis to obtain the understanding analysis results of the reference content, and the understanding analysis results of the reference content can be added to the knowledge base together with the reference content. Subsequently, content generation can be performed based on both the reference content and the understanding analysis results, which is conducive to increasing the amount of information based on the content generation process and improving the accuracy of content generation. For example, understanding analysis can be to summarize the reference content and obtain a summary of the text in the reference content.

[0263] In a possible implementation, during the process of parsing the media, the parsing progress of the media may be displayed on a knowledge base page of the knowledge base.

[0264] Specifically, refer to Fig.17 , Fig.17 An optional schematic diagram for displaying the parsing progress provided for the embodiment of the present disclosure may use a progress bar to display the parsing progress of the media, wherein after the reference content is added to the knowledge base, it may be displayed as a mark in the knowledge base page of the knowledge base through the third reference link 1211. The progress of the parsing progress may be displayed in the corresponding third reference link 1211, or displayed at other locations on the knowledge base page of the knowledge base, and the progress may be displayed as a specific percentage or a progress bar. By displaying the parsing progress of the media in the knowledge base page of the knowledge base, it is possible for the user to grasp the parsing progress of the media, thereby improving the management efficiency of the knowledge base.

[0265] In one possible implementation, when the medium is a web page, the medium is parsed to obtain reference content, specifically, the body and title of the web page are parsed to obtain a first parsing result; when it is determined that the first parsing result includes a path mark, the body and title of the first parsing result are parsed again to obtain a second parsing result; when it is detected that no content exists in the second parsing result, the text node in the web page is extracted as the second parsing result through the path rule; the second parsing result is cleaned of abnormal data to obtain reference content.

[0266] Among them, since web pages generally contain some information that is useless to the knowledge base, such as promotional information, style codes, scripts, etc., therefore, an attempt is made to parse the web page to obtain the first parsing result. On this basis, although the first parsing result already contains the text and title, the first parsing result is parsed again for the text and title, which can play a filtering role, further eliminate noise or unnecessary parts, and improve the accuracy and neatness of the parsing result. When it is detected that there is content in the second parsing result, the second parsing result can be directly cleaned for abnormal data to obtain reference content. When it is detected that there is no content in the second parsing result, there may be a parsing failure. At this time, the text node in the web page is extracted as the second parsing result through the path rule, which can ensure that there is content in the second parsing result. Finally, the parsed text and title may have abnormal situations. By cleaning the abnormal data of the second parsing result, the security of the reference content finally obtained can be improved. Through the above operations, accurate and pure text and title can be parsed from the web page as reference content.

[0267] Specifically, refer to Figures 18 to 21 , Figures 18 to 21 A schematic diagram of an optional process for parsing a web page provided in an embodiment of the present disclosure can be used to call a distiller tool to obtain a first parsing result, wherein, referring to Fig.18 ,First initialize the distiller rule, call the distiller tool for parsing, and when the distiller tool parses successfully, inject the first parsing result into reslma, where reslma is a pre-created data set used to store parsing results. When the distiller tool parses successfully, the body and title can be captured as the first parsing result through js and regular expressions.

[0268] Reference Fig.19 ,Then enter the filtering logic to detect whether the first parsing result includes the path tag (xpathPatern). If the path tag is included, parse the xpath rule, separate the text rule and the title rule, and parse each data from the first parsing result to see whether it is the text or the title through the text rule and the title rule. If it is the text or the title, return the text or the title through xpath to get the second parsing result and overwrite the updated reslma. Then the updated reslma can be passed into the custom parser to further optimize the parsing effect. The custom parser refers to a parser that can optimize the parsing effect of a web page by passing in the XPATH rule for a specific website. In addition, if the first parsing result does not include the path tag, the first parsing result can be directly passed into the custom parser to further optimize the parsing effect and obtain the final second parsing result.

[0269] Reference Fig. 20 ,Then perform a bottom-up processing, specifically, check whether there is content in the second parsing result. If there is no content in the second parsing result, the text node in the web page can be extracted through the path rule, and then determine whether the content formed by the extracted text node exceeds the line number threshold (for example, 50 lines, etc.). If it exceeds the line number threshold, the content of the extracted text node will be used as the second parsing result to overwrite the final reslma.

[0270] Finally, refer to Fig.21 ,The second parsing result is cleaned for abnormal data, mainly to filter out abnormal results, such as ,uniformly filtering domain names in the text and title, domain names, abnormal content, ,redundant titles, etc., and identifying blacklisted sites, for example, ,judging whether it is a blacklisted site based on the text, and if it is a blacklisted site, the final result is marked or directly ,eliminated.,After the abnormal data cleaning, the reference content is finally obtained and ,added to the knowledge base.

[0271] It is understandable that in addition to parsing out the main text and title, it is also possible to parse out pictures, videos, voice and other contents as reference content.

[0272] In addition, the above-mentioned process of parsing and cleaning the media can be completed on the terminal side. For example, after a user browses a web page or document on the terminal and adds it to the knowledge base, the corresponding algorithm can be called on the terminal side for parsing and cleaning, and the obtained reference content can be sent to the server and then added to the knowledge base. Alternatively, the terminal can send the domain name of the web page or document to the server, and the server will perform parsing and cleaning.

[0273] In a possible implementation, a knowledge base page of the knowledge base may be displayed, wherein the knowledge base page displays at least one third reference link, and in response to an operation of opening the third reference link, corresponding reference content is displayed.

[0274] Among them, the third reference link is used to access the reference content in the knowledge base. The knowledge base page can display multiple third reference links, each third reference link corresponds to each reference content one by one, and the third reference link can be displayed in the form of an icon. The icon can display part of the title, part of the text and a thumbnail, and can also display the type of media from which the corresponding reference content comes, such as from a document, a web page, etc. Fig. 22 , Fig. 22A schematic diagram of an optional operation for displaying reference content provided in an embodiment of the present disclosure, wherein a user can preview the reference content by opening the third reference link 1211, for example, the third reference link 1211 can be opened by double-clicking the third reference link 1211, wherein when previewing the reference content, the reference content obtained after parsing is displayed, for example, the reference content obtained after parsing the web page as mentioned above. By displaying the third reference link on the knowledge base page, it is convenient for the user to manage the reference content in the knowledge base, thereby improving the management efficiency of the knowledge base.

[0275] In one possible implementation, after the corresponding reference content is displayed by opening the third reference link, the modified reference content in the knowledge base can be updated in response to an operation of modifying the currently displayed reference content; and the updated reference content can be displayed in response to an operation of opening the third reference link again.

[0276] Among them, after the reference content is displayed, the user can modify it according to actual needs, such as modifying the text, title, or inserting pictures, videos, voices, etc. After the reference content is modified on the terminal, the corresponding modification operation can be sent to the server, and the server updates the original reference content in the knowledge base based on the modification operation. Alternatively, the terminal can generate updated reference content according to the modification operation, and then send the updated reference content to the server to overwrite the original reference content. By being able to modify the reference content after displaying the reference content, it is convenient for users to manage the reference content in the knowledge base and improve the management efficiency of the knowledge base.

[0277] In addition, in addition to modifying the reference content by displaying the reference content, it is also possible to modify the reference content by importing a file with the same name to overwrite the corresponding reference content.

[0278] In one possible implementation, referring to Fig.23 , Fig.23An optional schematic diagram of updating reference content provided in an embodiment of the present disclosure, the reference content may be divided into different data shards for storage, each reference content is assigned a corresponding version number when stored, such as "V1" shown in the figure, "Chunk_1, Chunk_2 and Chunk_N" are used to mark different data shards of the same reference content, and at the same time, each shard is assigned a corresponding database state, which is used to indicate the update progress of the data shard, for example, "pre-write" refers to the beginning of writing a new version of the reference content, and "submit" refers to the updated versions of all data shards of the reference content have been written, and the version number needs to be modified. When the update starts, the new version of the reference content is written first. At this time, the new version is divided into multiple data shards for writing (one-to-one correspondence with the old version of the data shards), and the database state of the data shards is modified to "pre-write". After all data shards are successfully written, the database state of the data shards is modified to "submit", and the original version number is modified, for example, to "V2", and finally the old version of the data shards are deleted.

[0279] Among them, when the database status is "pre-written", the corresponding data shard is not visible, that is, it cannot be recalled by the knowledge base. When the database status is "pre-written", the reference content can be recalled when the content is generated based on the full knowledge base, but because the version number has not been modified at this time, it is impossible to recall only the currently updated reference content (for example, preview the reference content). If a timeout occurs when the database status is changed to "submit", retry until the number of timeouts or the cumulative timeout duration reaches a preset threshold (for example, 10 times or 1 minute), then abandon this update. After abandoning this update, the database status may be successfully changed to "submit". At this time, the reference content can be recalled when the content is generated based on the full knowledge base, and only the old version of the content can be recalled when only the currently updated reference content is recalled. Similarly, if a timeout occurs when deleting the old version of the data shard, retry until the number of timeouts or the cumulative timeout duration reaches a preset threshold, then abandon this update. After abandoning this update, the database status may be successfully changed to "submitted". At this time, the reference content can be recalled when generating content based on the entire knowledge base. When recalling only the currently updated reference content, only the old version of the content can be recalled.

[0280] In a possible implementation, the knowledge base can be rebuilt and refreshed, referring to Fig.24 , Fig.24An optional schematic diagram of knowledge base reconstruction and refresh provided for the disclosed embodiment, the reconstruction and refresh adopts a dual-write scheme, the basis of the scheme is that the read client accesses the reference content under the corresponding index through the index alias. Specifically, the start time ts is first recorded, and when the write client writes the incremental reference content, a newly added index index_new is assigned to the incremental reference content, and the incremental reference content is written to the index index_new. At the same time, the incremental reference content is written to the index alias index_alia corresponding to the original index index_old, and marked as incremental data. Then, the reference content with a write time less than the start time ts is asynchronously rewritten to index_new. After the asynchronous process ends, the reference relationship between the index index_old and the index alias index_alia is deleted, and the reference relationship (atomic) between the index alias index_alia and the index index_new is constructed. At this time, the underlying data of the index index_new and the index alias index_alia is the same, and the user is unaware during the content generation process.

[0281] In one possible implementation, referring to Fig.25 , Fig.25 A schematic diagram of an optional operation for generating target content provided by the embodiment of the present disclosure is provided. When the knowledge base-based method is selected for content generation through the generation mode selection control, the target content corresponding to the first prompt information is generated in the session page 310 based on the knowledge base pre-created by the target account. Specifically, when the knowledge base-based method is selected for content generation through the generation mode selection control, the file import control 2520 is displayed; in response to the operation of importing a temporary file through the file import control 2520, the target content corresponding to the first prompt information is generated in the session page 310 based on the knowledge base pre-created by the target account and the temporary file. At the same time, the first reference link 316 corresponding to the reference content in the knowledge base and the fourth reference link 2540 corresponding to the temporary file are also displayed in the session page 310. Based on the knowledge base, the user can also import a temporary file, which contains reference content that does not exist in the knowledge base, and generate content by combining the temporary file with the knowledge base, thereby releasing the capacity of the knowledge base on the basis of making the content generation result more accurate.

[0282] After the user imports the temporary file, he can build a temporary knowledge base based on the temporary file. The temporary knowledge base has the same structure as the original knowledge base, that is, the temporary file is added to the temporary knowledge base after parsing. The temporary knowledge base can be used as a knowledge supplement for a user's content generation request. After completing the current content generation, the temporary knowledge base can be deleted.

[0283] The following is a specific example to illustrate the principle of the content generation method provided by the embodiment of the present disclosure.

[0284] Specifically, the content generation method provided by the embodiment of the present disclosure can be applied to a smart workbench that integrates reading, searching, and writing. The smart workbench can be run on a terminal. The smart workbench can specifically have functions such as knowledge base addition, knowledge base content preview, knowledge base search, and conversation. Among them, the knowledge base search can allow users to retrieve corresponding reference content for preview, and the conversation is realized based on the results of the knowledge base search.

[0285] First, the architecture of the intelligent workbench in the embodiment of the present disclosure is described as a whole. Fig.26 , Fig.26 The overall architecture diagram of the intelligent workbench provided by the embodiment of the present disclosure, wherein the reference content in the knowledge base can be added or deleted for management, and at the same time, the list of added users and the list of deleted users are recorded by maintaining the user database, which is convenient for subsequent tracing. The user uploads a temporary file to the media management module, and the media management module calls the parsing module to perform asynchronous parsing to obtain the reference content, and stores the media added to the knowledge base and the temporary file in the media database for storage. Whenever the reference content is parsed, the knowledge base and the temporary knowledge base are pre-written, wherein the knowledge base is used to write the reference content in the media added to the knowledge base, and the temporary knowledge base is used to write the reference content in the temporary file, and the database status of the reference content is updated synchronously. The intelligent workbench provides question-and-answer services and retrieval services, wherein the question-and-answer service includes temporary file mixed question-and-answer and knowledge base question-and-answer, the temporary file mixed question-and-answer adopts the recall method of the knowledge base and the temporary knowledge base, and the knowledge base question-and-answer adopts the recall method of the knowledge base. The retrieval service includes AI retrieval and ordinary retrieval, AI retrieval is to recall by extracting feature vectors, and ordinary retrieval is to recall by keyword matching, and both retrieval methods adopt the recall method of the knowledge base. The upstream can call the corresponding recall method through the scene ID (for example, it can include keyword recall, vector recall and mixed recall, mixed recall is keyword recall + vector recall), and the recall filtering conditions need to be passed in. Various recall strategies can be implemented, and the strategies can be combined (multi-way recall). The recalled data can be rearranged according to the strategy or using the rearrangement model.

[0286] The following is a detailed description of the process of adding content to the knowledge base in the intelligent workbench.

[0287] First, the user logs in to the target account through the smart workbench. The target account is associated with a pre-created knowledge base, which can be deployed on a terminal or a server.

[0288] Then, the user can use the function of adding reference content to the knowledge base in the smart workbench. Specifically, the user can upload media through the smart workbench to add reference content, where the media can be web pages, documents, audio or video, etc. Alternatively, the reference content in other media can be added to the knowledge base when browsing other media, for example, adding web pages to the knowledge base when browsing web pages in a browser, adding a shared content to the knowledge base in a social media application, etc.

[0289] Among them, the terminal can parse the media uploaded by the user to obtain reference content; then, when the knowledge base is deployed on the terminal, the terminal adds each data segment obtained by dividing the reference content to the knowledge base, and when the knowledge base is deployed on the server, the terminal sends the reference content to the server so that the server adds each data segment obtained by dividing the reference content to the knowledge base, or the terminal sends each data segment obtained by dividing the reference content to the server so that the server adds the data segment to the knowledge base.

[0290] In addition, the media can also be parsed by the server. Specifically, the terminal sends the media uploaded by the user to the server for the server to parse and obtain reference content. When the knowledge base is deployed on the server, the server adds each data slice obtained by dividing the reference content to the knowledge base, and when the knowledge base is deployed on the terminal, the server sends the reference content to the terminal for the terminal to add each data slice obtained by dividing the reference content to the knowledge base, or the server sends each data slice obtained by dividing the reference content to the terminal for the terminal to add the data slice to the knowledge base.

[0291] It should be noted that the analysis can eliminate abnormal and redundant content in the media, and only retain the key parts of the media. For example, the reference content can include the title, text, and pictures of the media. Adding each data segment of the reference content to the knowledge base is equivalent to adding the reference content to the knowledge base; the knowledge base will also record the reference content to which the data segment belongs and the version information of the data segment, for example, the data segment and the version information are stored in association.

[0292] Then, the data shards stored in the knowledge base can be updated. For example, when the data shard of the reference content doc1 is first added to the knowledge base, the version information corresponding to the data shard is set to v1. When the reference content doc1 is updated, the updated data shard of the reference content doc1 is added to the knowledge base again, and the version information corresponding to the updated data shard is set to v2. When the data shard is successfully added, the old version of the data shard corresponding to the reference content doc1 is deleted from the knowledge base, that is, the data shard with version information v1 corresponding to the reference content doc1 is deleted.

[0293] The following is a detailed description of the process of previewing and modifying reference content in the smart workbench.

[0294] First, after the user logs in to the target account through the smart workbench, he can use the knowledge base content preview function in the smart workbench. Specifically, the knowledge base page of the knowledge base is displayed in the smart workbench, and the third reference link corresponding to the reference content currently included in the knowledge base is displayed on the knowledge base page. After clicking the third reference link, the reference content can be displayed. The reference content can be a title, part of the text, and a picture, etc., among which the reference content can be suspended on the knowledge base page for display.

[0295] In addition, after previewing the reference content on the knowledge base page, users can modify the reference content and update it synchronously in the knowledge base after confirming the modification. In addition to modifying the reference content when previewing it, you can also upload a file with the same name in the knowledge base and parse it to get the updated reference content, thereby modifying the reference content.

[0296] The following is a detailed description of the process of performing knowledge base retrieval in the intelligent workbench.

[0297] Specifically, refer to Fig. 27 , Fig. 27 A schematic diagram of an optional architecture for knowledge base retrieval provided in an embodiment of the present disclosure.

[0298] The retrieval scheme of the knowledge base is divided into two parts: index establishment and retrieval. In order to cope with large-scale data storage, an extensible index management scheme is designed to facilitate subsequent expansion. At the same time, in order to improve the recall rate, two indexing methods are established, namely vector index and text index. For the recalled data, vectorization and word segmentation are performed based on the user's prompt information. In order to further improve the relevance of the shards, the sharding model is used for sharding and screening. Finally, in the index of the data shards, the similarity of the data shards is calculated through the sharding model, and the content of the large language model answer is indexed as a reference content in the knowledge base.

[0299] First, the process of index creation is described in detail below.

[0300] After adding each data fragment obtained by dividing the reference content to the knowledge base, indexing is performed in the following two ways:

[0301] Method 1: Extract features from data slices to obtain original vectors, and then process the original vectors based on the HNSW (Hierarchical Navigable Small Word) algorithm to establish vector indexes for data slices in the knowledge base. Fig.28 , Fig.28An optional schematic diagram of a quantization result provided in an embodiment of the present disclosure. In order to take into account the large-scale vector search speed, int8 quantization can be used to perform three-layer vectorization processing to obtain a vector index.

[0302] Method 2: By segmenting the data shards, a multi-field, multi-dimensional inverted index is established in the knowledge base, that is, a text index is obtained. Specifically, indexes can be established for titles, contents, tags, title pinyin, etc.

[0303] It should be noted that the reference Fig.29 , Fig.29 An optional schematic diagram of the organizational architecture provided for the embodiment of the present disclosure. In order to cope with the storage and retrieval of large-scale user data, an organizational structure based on multiple clusters and multiple indexes is designed. Clusters are dynamically allocated according to the growth of business data. The user account can be routed to different indexes (Index_0 to Index_199) according to the account ID (uid) of the user account, and then the sharded routing is performed by the account ID to ensure that the performance of writing and retrieval is within a controllable range to cope with the rapid growth of the business in the later stage.

[0304] Then, the retrieval process is described in detail below, refer to Fig.30 , Fig.30 An optional schematic diagram of a retrieval process provided for an embodiment of the present disclosure.

[0305] First, after the user logs in to the target account through the smart workbench, the user can use the knowledge base retrieval function in the smart workbench. Specifically, a content search page is displayed in the smart workbench. A text input box is displayed in the content search page, and the user can enter prompt information through the text input box.

[0306] Then, feature extraction is performed on the prompt information to obtain the query vector. Based on the HNSW algorithm and the query vector, the vector index of the knowledge base is retrieved to recall the data fragments with relevant semantics.

[0307] Then, the prompt information is segmented, and corresponding processing and search strategies are adopted for the segmentation results of different fields in the prompt information to recall text-related data fragments. For example, for the search of tags, a complete match search strategy will be selected. For another example, pinyin search will be used while searching based on the title. By integrating multiple strategies, BM25-based text retrieval can also be performed on the data fragments to further improve the recall rate.

[0308] Then, for the multi-way recalled data shards, it is necessary to calculate the final correlation score of each recalled data shard, and then reorder each recalled data shard based on the correlation score to obtain the final reordered sequence consisting of several data shards. The data shards can be reordered using a reordering model, which can be a multilingual reordering model fine-tuned based on the business corpus.

[0309] Then, according to the actual scenario, the first K data slices in the rearranged sequence can be used as the retrieval results of the prompt information, and these K data slices can be displayed in the intelligent workbench, where K is a positive integer, for example, K can be 10.

[0310] The process of conducting a conversation in the intelligent workbench is described in detail below.

[0311] Among them, the conversations of the smart workbench can be classified into two categories, one is question-and-answer conversations, and the other is creative conversations.

[0312] The following is an introduction to the process of question-and-answer conversation.

[0313] First, run the smart workbench on the terminal to provide content generation services, and log in to the target account on the smart workbench.

[0314] Then, the user opens the conversation page through the smart workbench and enters the first prompt information on the conversation page. In this scenario, the first prompt information is the user's question. A generation method selection control is provided in the conversation page. The generation method selection control is used to select content generation based on network search or content generation based on knowledge base.

[0315] Then, when the knowledge base-based method is selected for content generation through the generation method selection control, the target content can be generated based on the relevant reference content in the knowledge base. In this scenario, the target content is the answer to the user's question.

[0316] Specifically, refer to Fig.31 , Fig.31 An optional flow chart of outputting target content provided in an embodiment of the present disclosure, after the user inputs the first prompt information, the terminal sends the first prompt information to the server, the server extracts the reference content actively selected by the user from the knowledge base, or recalls the corresponding reference content from the knowledge base according to the first prompt information, splices the reference content with the first prompt information and inputs it into the large language model for inference, generates the target content and sends it to the terminal, so as to be displayed in the session page of the smart workbench.

[0317] Among them, the knowledge base and the large language model can also be deployed on the terminal side, and the above process of generating target content can be completed locally on the terminal.

[0318] The following describes the process of authoring a session.

[0319] The processing process of the creation session is similar to that of the question-and-answer session. The difference is that in the scenario of the creation session, the first prompt information is the user's creation requirements, which will not be repeated here.

[0320] In addition, after obtaining the target content through the conversation, the original text can be located in the reference content. Specifically, the target content can be parsed to obtain multiple answer points, where the answer points are different parts decomposed from the target content. The relevance between the answer points and the paragraphs in the reference content is calculated, and the paragraphs related to the answer points are determined in the reference content. The original text is located in the reference content according to the coordinate information corresponding to the paragraphs, so that users can quickly view the original reference content.

[0321] In addition, the smart workbench provided by the embodiment of the present disclosure is also provided with a note function. In response to the operation of adding content to the knowledge base in the conversation page, when the reference content generated by the target account through the conversation is added to the knowledge base, specifically, in response to the operation of adding content to the knowledge base in the conversation page, a new note control can be displayed; in response to the operation of triggering the new note control, a note page is displayed, wherein the note page includes a note list subpage and a note editing subpage displayed in parallel, and a first adding control is displayed in the note editing subpage; the target note of the new note control operation is highlighted in the note list subpage, and the reference content generated by the target account through the conversation is displayed in the note editing subpage corresponding to the target note; the reference content is edited in the note editing subpage, and in response to the operation of triggering the first adding control, the edited reference content is added to the knowledge base.

[0322] Specifically, refer to Fig.32 , Fig.32The schematic diagram of the operation of initiating editing of the generated target content provided in the embodiment of the present disclosure is that when the fourth add control 908 is triggered, instead of directly adding the generated reference content to the knowledge base, a new note control 3210 may be displayed first, wherein the new note control 3210 may be used to select a new note or an existing note. When the new note control 3210 is triggered, a note page 3220 is displayed. The note page 3220 includes a note list subpage 3221 and a note editing subpage 3222 displayed side by side. The first add control 3223 is displayed in the note editing subpage. If the new note method is selected, the target note "New Note" is highlighted in the note list subpage 3221 as the currently selected note, and the reference content is displayed in the note editing subpage 3222. Next, after editing the reference content in the note editing subpage, in response to triggering the operation of the first add control 3223, the edited reference content is added to the knowledge base. In this way, the target content can be edited and then added to the knowledge base by introducing the note function, allowing users to conveniently manage the knowledge base content.

[0323] Among them, the user can modify the reference content through the modification control 3224 in the note editing subpage 3222, such as bolding, italicizing, adding underline, changing font color, etc.

[0324] In one possible implementation, editing reference content in a note editing subpage may specifically be performed by displaying an editing control in response to selecting at least a portion of subcontent of the reference content in the note editing subpage, wherein the editing control is used to input editing requirements or to trigger a preset editing function; in response to an operation of triggering the editing control, generating editing content and a first insert control for inserting the editing content in the note editing subpage, and highlighting the editing content; in response to an operation of triggering the first insert control, adding the editing content to the reference content, and canceling the highlighting of the editing content.

[0325] Specifically, refer to Fig.33 , Fig.33 The operation diagram of the editing process of the generated target content provided by the embodiment of the present disclosure, after selecting the sub-content 3310, at least one of the writing control 3320 or the interpretation control 3330 can be displayed. When the writing control 3320 is triggered, the editing control 3340 is displayed, or the editing control 3340 can be directly displayed after selecting the sub-content 3310. The editing control 3340 may include one or more of a text input box for inputting editing requirements, an expansion control, an abbreviation control, and a translation control. After the editing control 3340 is triggered, refer to Fig.34 , Fig.34The operation diagram of the process of inserting edited content provided for the embodiment of the present disclosure, after the edited content 3410 is generated, the first insert control is displayed at the same time, wherein the highlighting method may be bolding or changing the color, etc., so that the user can distinguish the edited content, and the first insert control 3420 may be a replacement control, in which case the insertion method may be replacing the sub-content, and the first insert control 3420 may also be an insert control, in which case the insertion method may be replacing the sub-content or adding to the end of the sub-content. In addition, a control 3430 for regenerating the edited content and a control 3440 for discarding the edited content may also be displayed. After triggering the first insert control 3420, the edited content 3410 obtained by AI editing may be inserted into the reference content and the highlight may be canceled, so that the edited content is integrated with the original reference content. Through the above operations, the reference content can be quickly edited in the note editing subpage, thereby improving the efficiency of human-computer interaction.

[0326] In one possible implementation, editing reference content in a note editing subpage may specifically be responsive to selecting at least a portion of subcontent of the reference content in the note editing subpage, displaying an interpretation control; responsive to an operation that triggers the interpretation control, displaying an interpretation page, displaying the interpretation content of the subcontent and a second add control in the interpretation page; responsive to an operation that triggers the second add control, adding the interpretation content and a second insert control for inserting the interpretation content in the note editing subpage, and highlighting the interpretation content; responsive to the operation that triggers the second insert control, adding the interpretation content to the reference content, and canceling the highlighting of the interpretation content.

[0327] Specifically, refer again to Fig.33 , assuming that when the interpretation control 3330 is triggered, the interpretation page is displayed, refer to Fig.35 , refer to Fig.35 , Fig.35 Schematic diagram of the operation of the interpretation process provided by the embodiment of the present disclosure. The interpretation page 3510 can be displayed in parallel with the note editing subpage 3222. The interpretation page 3510 displays the interpretation content 3520 and the second adding control 3530. The interpretation content 3520 is used to describe the selected subcontent 3310 in detail. After the second adding control 3530 is triggered, the interpretation content 3520 can be added to the reference content for highlighting. The specific principle is the same as Fig.34 The highlighting principle shown is consistent with that shown, and a second insert control is displayed. The second insert control is consistent with the principle of the first insert control, and no further description is given here. Add the interpretation content to the reference content and cancel the highlighting of the interpretation content. Through the above operation, the reference content can be quickly edited in the note editing subpage, thereby improving the efficiency of human-computer interaction.

[0328] In addition, after triggering the second adding control 3530, the interpretation content 3520 can also be directly added to the knowledge base.

[0329] In addition, refer to Fig.36 , Fig.36 A schematic diagram of an optional interface of the smart workbench provided in the embodiment of the present disclosure, wherein the smart workbench can simultaneously provide a knowledge base entry 3610 and a note entry 3620. The knowledge base page can be quickly displayed through the knowledge base entry 3610, and the note page can be quickly displayed through the note entry 3620, thereby realizing functional integration. The user can directly display the note page 3220 by triggering the note entry 3620, create a new note by triggering the new note control 3630 in the note page 3220, manually enter content in the note editing subpage 3222 corresponding to the new note, or automatically generate content through AI, and after entering content or generating content in the note editing subpage 3222, the user can use Figure 33 to Figure 35 The content can be edited in a new way and the content of the newly created note can be added to the knowledge base as reference content, thereby effectively improving the flexibility of the reference content in the knowledge base.

[0330] The intelligent workbench provided by the embodiment of the present disclosure can generate target content that is more refined and more in line with the private knowledge field of the target account, making the target content more focused and professional, thereby effectively improving the accuracy of the target content.

[0331] It is to be understood that, although the steps in the above-mentioned various flow charts are shown in sequence according to the indication of the arrows, these steps are not necessarily performed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in the present embodiment, the execution of these steps does not have a strict order restriction, and these steps can be performed in other orders. Moreover, at least a portion of the steps in the above-mentioned flow charts can include a plurality of steps or a plurality of stages, and these steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of the steps or stages in other steps or other steps.

[0332] Reference Fig.37 , Fig.37 This is an optional structural diagram of a content generation device provided in an embodiment of the present disclosure. The content generation device 3700 includes:

[0333] The acquisition module 3701 is used to run the content generation service, wherein the content generation service is logged in with the target account;

[0334] The generation module 3702 is used to obtain the first prompt information input into the content generation service. When the knowledge base-based method is selected to generate content in the content generation service, the target content corresponding to the first prompt information is generated in the content generation service based on the knowledge base pre-created by the target account.

[0335] Further, the first prompt information is input through a conversation page of the content generation service, and the conversation page displays a generation mode selection control. The generation module 3702 is specifically used for:

[0336] When the knowledge base-based method is selected for content generation through the generation method selection control, the target content corresponding to the first prompt information is generated in the session page based on the knowledge base pre-created by the target account.

[0337] Furthermore, the generation module 3702 is also used for:

[0338] displaying at least one first reference link in the session page;

[0339] The first reference link is used to access reference content in the knowledge base for generating target content.

[0340] Furthermore, the above-mentioned content generation device further includes a regeneration module 3702 (not shown in the figure), and the regeneration module 3702 is specifically used for:

[0341] In response to an operation of selecting at least one first reference link in the conversation page, marking the selected first reference link in the conversation page;

[0342] In response to the operation of regenerating the target content in the conversation page, the target content is regenerated in the conversation page based on the reference content corresponding to the selected first reference link.

[0343] Furthermore, the generation module 3702 is specifically used for:

[0344] When the knowledge base-based method is selected for content generation through the generation method selection control, the file import control is displayed;

[0345] In response to the operation of importing a temporary file through the file import control, target content corresponding to the first prompt information is generated in the session page based on the knowledge base and the temporary file pre-created by the target account.

[0346] Furthermore, the generation module 3702 is specifically used for:

[0347] When the knowledge base-based method is selected for content generation through the generation method selection control, the knowledge base page is displayed, wherein the knowledge base page is used to display reference content in the knowledge base pre-created by the target account;

[0348] In response to an operation of selecting at least one reference content in the knowledge base page, target content corresponding to the first prompt information is generated in the conversation page based on the selected reference content.

[0349] Furthermore, the above content generation device further includes a first updating module (not shown in the figure), and the first updating module is specifically used for:

[0350] In response to the operation of adding content to the knowledge base, the reference content browsed or processed by the target account is added to the knowledge base.

[0351] Furthermore, the first update module is specifically used for:

[0352] Display the content import page of the knowledge base;

[0353] In response to an operation of adding content to the knowledge base in the content import page, the reference content imported by the target account is added to the knowledge base.

[0354] Furthermore, the first update module is specifically used for:

[0355] Displaying a conversation page of the content generation service, and acquiring second prompt information input in the conversation page;

[0356] When generating content in the conversation page, generating reference content corresponding to the second prompt information in the conversation page;

[0357] In response to an operation of adding content to the knowledge base in the session page, the reference content generated by the target account through the session is added to the knowledge base.

[0358] Furthermore, the first update module is specifically used for:

[0359] In response to an operation of adding content to the knowledge base in the conversation page, displaying a new note control;

[0360] In response to the operation of triggering a new note control, displaying a note page, wherein the note page includes a note list subpage and a note editing subpage displayed in parallel, and a first adding control is displayed in the note editing subpage;

[0361] The target note of the new note control operation is highlighted in the note list subpage, and the reference content generated by the target account through the conversation is displayed in the note editing subpage corresponding to the target note;

[0362] The reference content is edited in the note editing subpage, and in response to triggering the operation of the first adding control, the edited reference content is added to the knowledge base.

[0363] Furthermore, the first update module is specifically used for:

[0364] In response to selecting at least a portion of the sub-content of the reference content in the note editing sub-page, displaying an editing control, wherein the editing control is used to input an editing requirement or to trigger a preset editing function;

[0365] In response to the operation of triggering the edit control, generating edit content and a first insert control for inserting the edit content in the note editing subpage, and highlighting the edit content;

[0366] In response to the operation of triggering the first insert control, the edited content is added to the reference content, and the highlighting of the edited content is canceled.

[0367] Furthermore, the first update module is specifically used for:

[0368] In response to selecting at least a portion of the sub-content of the reference content in the note editing sub-page, displaying an interpretation control;

[0369] In response to the operation of triggering the interpretation control, displaying the interpretation page, displaying the interpretation content of the sub-content and the second adding control in the interpretation page;

[0370] In response to the operation of triggering the second adding control, adding the interpretation content and the second inserting control for inserting the interpretation content in the note editing subpage, and highlighting the interpretation content;

[0371] In response to the operation of triggering the second insert control, the interpreted content is added to the reference content, and the highlighting of the interpreted content is cancelled.

[0372] Furthermore, the first update module is specifically used for:

[0373] Display the media pages browsed by the target account;

[0374] In response to an operation of adding content to the knowledge base in the media page, the reference content in the media page is added to the knowledge base.

[0375] Furthermore, the first update module is specifically used for:

[0376] Displaying a session page of the content generation service, when a network search-based method is selected in the session page to generate content, displaying at least one second reference link after the content is generated in the session page;

[0377] In response to the operation of opening the second reference link, a media page browsed by the target account is displayed.

[0378] Furthermore, the first update module is specifically used for:

[0379] In response to an operation of adding content to the knowledge base, obtaining media browsed or processed by the target account;

[0380] The media is parsed to obtain reference content, and the reference content is added to the knowledge base.

[0381] Furthermore, the first update module is also used for:

[0382] During the process of parsing the media, the parsing progress of the media is displayed on the knowledge base page of the knowledge base.

[0383] Furthermore, the medium is a web page, and the first updating module is specifically used for:

[0384] Parsing the body and title of the web page to obtain a first parsing result;

[0385] When it is determined that the first parsing result includes the path mark, the first parsing result is parsed again for the body and title to obtain a second parsing result;

[0386] When it is detected that no content exists in the second parsing result, a text node in the web page is extracted as the second parsing result by using the path rule;

[0387] The second analysis result is cleaned of abnormal data to obtain reference content.

[0388] Furthermore, the above-mentioned content generation device further includes a display module (not shown in the figure), and the display module is specifically used for:

[0389] Displaying a knowledge base page of the knowledge base, wherein the knowledge base page displays at least one third reference link, and the third reference link is used to access reference content in the knowledge base;

[0390] In response to an operation of opening the third reference link, the corresponding reference content is displayed.

[0391] Furthermore, the above-mentioned content generation device further includes a second updating module (not shown in the figure), and the updating module is specifically used for:

[0392] In response to an operation of modifying the currently displayed reference content, updating the modified reference content in the knowledge base;

[0393] In response to an operation of opening the third reference link again, the updated reference content is displayed.

[0394] The above-mentioned content generation device 3700 and content generation method are based on the same inventive concept. By running the content generation service logged in to the target account, the first prompt information input into the content generation service is obtained. When the knowledge base-based method is selected in the content generation service to generate content, the target content corresponding to the first prompt information is generated in the content generation service based on the knowledge base. Since the knowledge base is pre-created by the target account, the target account can store private professional knowledge through the knowledge base. Based on the knowledge base, it can generate more refined target content that is more in line with the private knowledge field of the target account, making the target content more focused and professional, thereby effectively improving the accuracy of the target content.

[0395] The electronic device for executing the above content generation method provided in the embodiment of the present disclosure may be a terminal. Fig.38 , Fig.38 This is a partial structural block diagram of a terminal provided in an embodiment of the present disclosure, and the terminal includes: a camera assembly 3810, a first memory 3820, an input unit 3830, a display unit 3840, a sensor 3850, an audio circuit 3860, a wireless fidelity (WiFi) module 3870, a first processor 3880, and a first power supply 3890. Those skilled in the art can understand that Fig.38 The terminal structure shown in the figure does not constitute a limitation on the terminal, and may include more or less components than shown in the figure, or combine certain components, or arrange the components differently.

[0396] The camera assembly 3810 can be used to capture images or videos. Optionally, the camera assembly 3810 includes a front camera and a rear camera. Typically, the front camera is disposed on the front panel of the terminal, and the rear camera is disposed on the back of the terminal. In some embodiments, there are at least two rear cameras, which are any one of a main camera, a depth of field camera, a wide-angle camera, and a telephoto camera, so as to realize the fusion of the main camera and the depth of field camera to realize the background blur function, the fusion of the main camera and the wide-angle camera to realize the panoramic shooting and VR (Virtual Reality) shooting function or other fusion shooting functions.

[0397] The first memory 3820 may be used to store software programs and modules. The first processor 3880 executes various functional applications and data processing of the terminal by running the software programs and modules stored in the first memory 3820 .

[0398] The input unit 3830 may be used to receive input digital or character information and generate key signal input related to the terminal's settings and function control. Specifically, the input unit 3830 may include a touch panel 3831 and other input devices 3832 .

[0399] The display unit 3840 may be used to display input information or provided information and various menus of the terminal. The display unit 3840 may include a display panel 3841.

[0400] The audio circuit 3860, the speaker 3861, and the microphone 3862 may provide an audio interface.

[0401] The first power source 3890 may be alternating current, direct current, a disposable battery, or a rechargeable battery.

[0402] The number of sensors 3850 may be one or more, and the one or more sensors 3850 include but are not limited to: acceleration sensors, gyroscope sensors, pressure sensors, optical sensors, etc. Among them:

[0403] The acceleration sensor can detect the magnitude of acceleration on the three coordinate axes of the coordinate system established by the terminal. For example, the acceleration sensor can be used to detect the components of gravity acceleration on the three coordinate axes. The first processor 3880 can control the display unit 3840 to display the user interface in a horizontal view or a vertical view according to the gravity acceleration signal collected by the acceleration sensor. The acceleration sensor can also be used for collecting game or user motion data.

[0404] The gyroscope sensor can detect the body direction and rotation angle of the terminal, and the gyroscope sensor can cooperate with the acceleration sensor to collect the user's 3D actions on the terminal. The first processor 3880 can implement the following functions based on the data collected by the gyroscope sensor: motion sensing (such as changing the UI according to the user's tilt operation), image stabilization during shooting, game control, and inertial navigation.

[0405] The pressure sensor can be set in the side frame of the terminal and / or the lower layer of the display unit 3840. When the pressure sensor is set in the side frame of the terminal, the user's holding signal of the terminal can be detected, and the first processor 3880 performs left and right hand recognition or shortcut operation according to the holding signal collected by the pressure sensor. When the pressure sensor is set in the lower layer of the display unit 3840, the first processor 3880 controls the operability controls on the UI interface according to the user's pressure operation on the display unit 3840. The operability control includes at least one of a button control, a scroll bar control, an icon control, and a menu control.

[0406] The optical sensor is used to collect the ambient light intensity. In one embodiment, the first processor 3880 can control the display brightness of the display unit 3840 according to the ambient light intensity collected by the optical sensor. Specifically, when the ambient light intensity is high, the display brightness of the display unit 3840 is increased; when the ambient light intensity is low, the display brightness of the display unit 3840 is decreased. In another embodiment, the first processor 3880 can also dynamically adjust the shooting parameters of the camera assembly 3810 according to the ambient light intensity collected by the optical sensor.

[0407] In this embodiment, the first processor 3880 included in the terminal can execute the content generating method of the previous embodiment.

[0408] The electronic device for executing the above-mentioned content generation method provided in the embodiment of the present disclosure may also be a server. Fig.39 , Fig.39 Partial structural block diagram of a server provided in an embodiment of the present disclosure. The server may have relatively large differences due to different configurations or performances, and may include one or more second processors 3910 and a second memory 3930, and one or more storage media 3940 (e.g., one or more mass storage devices) storing application programs 3943 or data 3942. Among them, the second memory 3930 and the storage medium 3940 may be short-term storage or permanent storage. The program stored in the storage medium 3940 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations on the server. Furthermore, the second processor 3910 may be configured to communicate with the storage medium 3940 and execute a series of instruction operations in the storage medium 3940 on the server.

[0409] The server may also include one or more second power supplies 3920, one or more wired or wireless network interfaces 3950, one or more input and output interfaces 3960, and / or one or more operating systems 3941, such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, etc.

[0410] The second processor 3910 in the server may be configured to execute the content generating method.

[0411] The embodiments of the present disclosure further provide a computer-readable storage medium, which is used to store a computer program, and the computer program is used to execute the content generation method of each of the above embodiments.

[0412] The embodiment of the present disclosure also provides a computer program product, which includes a computer program stored in a computer-readable storage medium. A processor of a computer device reads the computer program from the computer-readable storage medium, and the processor executes the computer program, so that the computer device executes the above-mentioned content generation method.

[0413] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present disclosure described herein can, for example, be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0414] It should be understood that in the present disclosure, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0415] It should be understood that in the description of the embodiments of the present disclosure, the meaning of multiple (or multiple items) is more than two, greater than, less than, exceed, etc. are understood to not include the number, and above, below, within, etc. are understood to include the number.

[0416] In the several embodiments provided in the present disclosure, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0417] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0418] In addition, each functional unit in each embodiment of the present disclosure may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0419] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present disclosure is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the various embodiments of the present disclosure. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store program codes.

[0420] It should also be understood that the various implementations provided in the embodiments of the present disclosure can be combined arbitrarily to achieve different technical effects.

[0421] The above is a specific description of the preferred implementation of the present disclosure, but the present disclosure is not limited to the above-mentioned implementation mode. Technical personnel familiar with the field can also make various equivalent deformations or substitutions under the shared conditions without violating the spirit of the present disclosure. These equivalent deformations or substitutions are all included in the scope defined by the claims of the present disclosure.

Claims

1. A content generation method, characterized in that: include: Running a content generation service, wherein the content generation service is logged into a target account; A first prompt information input into the content generation service is obtained, and when a knowledge base-based method is selected in the content generation service to generate content, target content corresponding to the first prompt information is generated in the content generation service based on the knowledge base pre-created by the target account.

2. The content generation method according to claim 1, characterized in that: The first prompt information is input through a session page of the content generation service, and the session page displays a generation mode selection control. When a knowledge base-based mode is selected in the content generation service to generate content, the target content corresponding to the first prompt information is generated in the content generation service based on the knowledge base pre-created by the target account, including: When the knowledge base-based method is selected to generate content through the generation method selection control, the target content corresponding to the first prompt information is generated in the session page based on the knowledge base pre-created by the target account.

3. The content generation method according to claim 2, characterized in that: The content generation method further includes: displaying at least one first reference link in the session page; The first reference link is used to access reference content in the knowledge base used to generate the target content.

4. The content generation method according to claim 3, characterized in that: The content generation method further includes: In response to an operation of selecting at least one of the first reference links in the conversation page, marking the selected first reference link in the conversation page; In response to the operation of regenerating the target content in the conversation page, the target content is regenerated in the conversation page based on the reference content corresponding to the selected first reference link.

5. The content generation method according to claim 2, characterized in that: When the knowledge base-based method is selected to generate content through the generation method selection control, the target content corresponding to the first prompt information is generated in the session page based on the knowledge base pre-created by the target account, including: When the knowledge base-based method is selected to generate content through the generation method selection control, a file import control is displayed; In response to the operation of importing a temporary file through the file import control, target content corresponding to the first prompt information is generated in the session page based on the knowledge base pre-created by the target account and the temporary file.

6. The content generation method according to claim 2, characterized in that: When the knowledge base-based method is selected to generate content through the generation method selection control, the target content corresponding to the first prompt information is generated in the session page based on the knowledge base pre-created by the target account, including: When the knowledge base-based method is selected for content generation through the generation method selection control, a knowledge base page is displayed, wherein the knowledge base page is used to display reference content in the knowledge base pre-created by the target account; In response to an operation of selecting at least one reference content in the knowledge base page, a target content corresponding to the first prompt information is generated in the conversation page based on the selected reference content.

7. The content generation method according to claim 1, characterized in that: Before acquiring the first prompt information input to the content generation service, the content generation method further includes: In response to an operation of adding content to the knowledge base, the reference content browsed or processed by the target account is added to the knowledge base.

8. The content generation method according to claim 7, characterized in that: In response to the operation of adding content to the knowledge base, adding the reference content browsed or processed by the target account to the knowledge base includes: Displaying a content import page of the knowledge base, and in response to an operation of adding content to the knowledge base on the content import page, adding the reference content imported by the target account to the knowledge base; Alternatively, displaying a session page of the content generation service, acquiring second prompt information input in the session page, generating reference content corresponding to the second prompt information in the session page when generating content in the session page, and in response to an operation of adding content to the knowledge base in the session page, adding the reference content generated by the target account through the session to the knowledge base; Alternatively, a media page browsed by the target account is displayed, and in response to an operation of adding content to the knowledge base in the media page, reference content in the media page is added to the knowledge base.

9. The content generation method according to claim 8, characterized in that: In response to the operation of adding content to the knowledge base in the conversation page, adding the reference content generated by the target account through the conversation to the knowledge base includes: In response to an operation of adding content to the knowledge base in the conversation page, displaying a new note control; In response to the operation of triggering the new note control, displaying a note page, wherein the note page includes a note list subpage and a note editing subpage displayed in parallel, and the note editing subpage displays a first adding control; Highlighting the target note of the new note control operation in the note list subpage, and displaying the reference content generated by the target account through the conversation in the note editing subpage corresponding to the target note; The reference content is edited in the note editing subpage, and in response to triggering the operation of the first adding control, the edited reference content is added to the knowledge base.

10. The content generation method according to claim 9, characterized in that: The editing of the reference content in the note editing subpage includes: In response to selecting at least a portion of the sub-content of the reference content in the note editing sub-page, displaying an editing control, wherein the editing control is used to input an editing requirement or to trigger a preset editing function; In response to the operation of triggering the editing control, generating editing content and a first insert control for inserting the editing content in the note editing subpage, and highlighting the editing content; In response to the operation of triggering the first insert control, the edited content is added to the reference content, and the highlighting of the edited content is canceled.

11. The content generation method according to claim 9, characterized in that: The editing of the reference content in the note editing subpage includes: In response to selecting at least a portion of the sub-content of the reference content in the note editing sub-page, displaying an interpretation control; In response to the operation of triggering the interpretation control, displaying an interpretation page, displaying the interpretation content of the sub-content and a second adding control in the interpretation page; In response to the operation of triggering the second adding control, adding the interpretation content and a second inserting control for inserting the interpretation content in the note editing subpage, and highlighting the interpretation content; In response to the operation of triggering the second insertion control, the interpretation content is added to the reference content, and the highlighting of the interpretation content is canceled.

12. The content generation method according to claim 8, characterized in that: The media page displayed that the target account browses includes: Displaying a session page of the content generation service, and when a network search-based method is selected in the session page to generate content, displaying at least one second reference link after generating content in the session page; In response to the operation of opening the second reference link, a media page browsed by the target account is displayed.

13. The content generation method according to claim 7, characterized in that: In response to the operation of adding content to the knowledge base, adding the reference content browsed or processed by the target account to the knowledge base includes: In response to an operation of adding content to the knowledge base, obtaining media browsed or processed by the target account; The medium is parsed to obtain reference content, and the reference content is added to the knowledge base.

14. The content generation method according to claim 13, characterized in that: The content generation method further includes: During the process of parsing the medium, the parsing progress of the medium is displayed on a knowledge base page of the knowledge base.

15. The content generation method according to claim 13, characterized in that: The medium is a web page, and the parsing of the medium to obtain reference content includes: Parsing the body and title of the webpage to obtain a first parsing result; When it is determined that the first parsing result includes a path mark, the first parsing result is parsed again for the body and title to obtain a second parsing result; When it is detected that no content exists in the second parsing result, extracting a text node in the web page as the second parsing result by using a path rule; The second analysis result is cleaned of abnormal data to obtain reference content.

16. The content generation method according to claim 1, characterized in that: The content generation method further includes: Displaying a knowledge base page of the knowledge base, wherein the knowledge base page displays at least one third reference link, and the third reference link is used to access the reference content in the knowledge base; In response to an operation of opening the third reference link, the corresponding reference content is displayed.

17. The content generation method according to claim 16, characterized in that: The content generation method further includes: In response to an operation of modifying the currently displayed reference content, updating the modified reference content in the knowledge base; In response to an operation of reopening the third reference link, the updated reference content is displayed.

18. A content generating device, characterized in that: include: An operation module, used to operate a content generation service, wherein the content generation service is logged in with a target account; A generation module is used to obtain first prompt information input into the content generation service, and when a knowledge base-based method is selected in the content generation service to generate content, the target content corresponding to the first prompt information is generated in the content generation service based on the knowledge base pre-created by the target account.

19. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the content generating method according to any one of claims 1 to 17 is implemented.

20. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the content generation method according to any one of claims 1 to 17 is implemented.

21. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the content generation method according to any one of claims 1 to 17 is implemented.

Images