Content generation method and device, electronic equipment, medium and program product

Abstract

The invention provides a content generation method and device, electronic equipment, a medium and a program product, relates to the technical field of artificial intelligence, in particular to the field of large language models and man-machine interaction, and can be used for application scenes such as generative search, document intelligent editing, intelligent assistants, virtual assistants and intelligent e-commerce. According to the implementation scheme, query information of a user is obtained; structured understanding is carried out on the query information to obtain a content generation strategy corresponding to the query information, and the content generation strategy comprises two or more sub-problems associated with the query information; providing a user interaction interface for outputting a content generation strategy and receiving an adjustment made by a user to the content generation strategy; and generating reply content corresponding to the query information based on the adjusted content generation strategy.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, in particular to large language models and human-computer interaction, and can be used in application scenarios such as generative search, intelligent document editing, intelligent assistants, virtual assistants, and intelligent e-commerce. It specifically relates to an artificial intelligence-based content generation method, an artificial intelligence-based content generation device, an electronic device, a computer-readable storage medium, and a computer program product. Background Art

[0002] Artificial intelligence is a discipline that studies how to use computers to simulate certain human thought processes and intelligent behaviors (such as learning, reasoning, thinking, planning, etc.). It includes both hardware-level and software-level technologies. Artificial intelligence hardware technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, and big data processing; artificial intelligence software technologies mainly include computer vision technology, speech recognition technology, natural language processing technology, as well as machine learning / deep learning, big data processing technology, knowledge graph technology, and other major directions.

[0003] Human-computer interaction refers to the information exchange process between people and computers in a certain interactive way to complete a certain task. The human-computer interaction system can be a variety of machines, or it can be a computerized system and software. The human-computer interaction interface usually refers to the part visible to the user. The user communicates with the system through the human-computer interaction interface and operates it. Improving the intelligence level of artificial intelligence through human-computer interaction is still one of the hot topics.

[0004] The methods described in this section are not necessarily methods that have been previously conceived or employed. Unless otherwise indicated, it should not be assumed that any method described in this section is considered to be prior art simply because it is included in this section. Similarly, unless otherwise indicated, the issues mentioned in this section should not be considered to have been recognized in any prior art. Summary of the invention

[0005] The present disclosure provides an artificial intelligence-based content generation method, an artificial intelligence-based content generation device, an electronic device, a computer-readable storage medium, and a computer program product.

[0006] According to one aspect of the present disclosure, there is provided an artificial intelligence-based content generation method, comprising: obtaining query information of a user; performing a structured understanding of the query information to obtain a content generation strategy corresponding to the query information, wherein the content generation strategy comprises two or more sub-questions associated with the query information; providing a user interaction interface for outputting the content generation strategy and receiving adjustments made by the user to the content generation strategy; and generating reply content corresponding to the query information based on the adjusted content generation strategy.

[0007] According to another aspect of the present disclosure, there is provided an artificial intelligence-based content generation device, comprising: an information acquisition module, configured to acquire query information of a user; a strategy generation module, configured to perform a structured understanding of the query information to obtain a content generation strategy corresponding to the query information, wherein the content generation strategy comprises two or more sub-questions associated with the query information; an interaction module, configured to provide a user interaction interface for outputting the content generation strategy and receiving adjustments made by the user to the content generation strategy; and a reply module, configured to generate reply content corresponding to the query information based on the adjusted content generation strategy.

[0008] According to another aspect of the present disclosure, an electronic device is provided, comprising at least one processor; and a memory communicatively connected to the at least one processor, wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method as described above.

[0009] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause a computer to execute the method as described above.

[0010] According to another aspect of the present disclosure, a computer program product is provided, comprising a computer program, wherein the computer program implements the method described above when executed by a processor.

[0011] According to one or more embodiments of the present disclosure, artificial intelligence-generated content can be made more in line with user needs.

[0012] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The accompanying drawings exemplarily illustrate the embodiments and constitute a part of the specification, and together with the text description of the specification, are used to explain the exemplary implementation of the embodiments. The embodiments shown are for illustrative purposes only and do not limit the scope of the claims. In all drawings, the same reference numerals refer to similar but not necessarily identical elements.

[0014] Figure 1 A schematic diagram showing an exemplary system in which the various methods described herein may be implemented according to an embodiment of the present disclosure;

[0015] Figure 2 A flowchart of a method for generating content based on artificial intelligence according to an embodiment of the present disclosure is shown;

[0016] Figure 3 A flow chart of obtaining user query information according to an embodiment of the present disclosure is shown;

[0017] Figure 4 A schematic diagram of an artificial intelligence-based content generation architecture according to an embodiment of the present disclosure is shown;

[0018] Figure 5 A structural block diagram of an artificial intelligence-based content generation device according to an embodiment of the present disclosure is shown;

[0019] Figure 6 A structural block diagram of an artificial intelligence-based content generation device according to an embodiment of the present disclosure is shown;

[0020] Figure 7 A structural block diagram of an exemplary electronic device that can be used to implement the embodiments of the present disclosure is shown. DETAILED DESCRIPTION

[0021] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0022] In the present disclosure, unless otherwise specified, the use of the terms "first", "second", etc. to describe various elements is not intended to limit the positional relationship, temporal relationship, or importance relationship of these elements, and such terms are only used to distinguish one element from another element. In some examples, the first element and the second element may refer to the same instance of the element, and in some cases, based on the description of the context, they may also refer to different instances.

[0023] The terms used in the description of various examples in this disclosure are only for the purpose of describing specific examples and are not intended to be limiting. Unless the context clearly indicates otherwise, if the number of elements is not specifically limited, the element can be one or more. In addition, the term "and / or" used in this disclosure covers any one of the listed items and all possible combinations.

[0024] The embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings.

[0025] Figure 1 FIG. 1 is a schematic diagram of an exemplary system 100 in which various methods and apparatuses described herein may be implemented according to an embodiment of the present disclosure. Figure 1, the system 100 includes one or more client devices 101, 102, 103, 104, 105, and 106, a server 120, and one or more communication networks 110 coupling the one or more client devices to the server 120. The client devices 101, 102, 103, 104, 105, and 106 may be configured to execute one or more applications.

[0026] In an embodiment of the present disclosure, the server 120 may run one or more services or software applications that enable execution of an artificial intelligence-based content generation method.

[0027] In some embodiments, server 120 may also provide other services or software applications, which may include non-virtualized environments and virtualized environments. In some embodiments, these services may be provided as web-based services or cloud services, such as provided to users of client devices 101, 102, 103, 104, 105, and / or 106 under a software as a service (SaaS) model.

[0028] exist Figure 1 In the configuration shown, the server 120 may include one or more components that implement the functions performed by the server 120. These components may include software components, hardware components, or a combination thereof that can be executed by one or more processors. Users operating client devices 101, 102, 103, 104, 105, and / or 106 may in turn utilize one or more client applications to interact with the server 120 to utilize the services provided by these components. It should be understood that a variety of different system configurations are possible, which may differ from the system 100. Therefore, Figure 1 is one example of a system for implementing the various methods described herein and is not intended to be limiting.

[0029] A user may use client devices 101, 102, 103, 104, 105, and / or 106 to provide query information and make adjustments to content generation strategies. The client device may provide an interface that enables a user of the client device to interact with the client device. The client device may also output information to the user via the interface. Figure 1 Only six client devices are depicted, but one skilled in the art will appreciate that the present disclosure may support any number of client devices.

[0030] Client devices 101, 102, 103, 104, 105 and / or 106 may include various types of computer devices, such as portable handheld devices, general-purpose computers (such as personal computers and laptop computers), workstation computers, wearable devices, smart screen devices, self-service terminal devices, service robots, game systems, thin clients, various messaging devices, sensors or other sensing devices, etc. These computer devices may run various types and versions of software applications and operating systems, such as MICROSOFT Windows, APPLE iOS, UNIX-like operating systems, Linux or Linux-like operating systems (such as GOOGLE Chrome OS); or include various mobile operating systems, such as MICROSOFT Windows Mobile OS, iOS, Windows Phone, Android. Portable handheld devices may include cellular phones, smart phones, tablet computers, personal digital assistants (PDAs), etc. Wearable devices may include head-mounted displays (such as smart glasses) and other devices. Game systems may include various handheld game devices, Internet-enabled game devices, etc. Client devices are capable of executing various different applications, such as various Internet-related applications, communication applications (such as email applications), short message service (SMS) applications, and may use various communication protocols.

[0031] The network 110 may be any type of network known to those skilled in the art that may support data communications using any of a variety of available protocols, including but not limited to TCP / IP, SNA, IPX, etc. By way of example only, the one or more networks 110 may be a local area network (LAN), an Ethernet-based network, a token ring, a wide area network (WAN), the Internet, a virtual network, a virtual private network (VPN), an intranet, an extranet, a blockchain network, a public switched telephone network (PSTN), an infrared network, a wireless network (e.g., Bluetooth, WIFI), and / or any combination of these and / or other networks.

[0032] Server 120 may include one or more general purpose computers, dedicated server computers (e.g., PC (personal computer) servers, UNIX servers, mid-range servers), blade servers, mainframe computers, server clusters, or any other suitable arrangement and / or combination. Server 120 may include one or more virtual machines running virtual operating systems, or other computing architectures involving virtualization (e.g., one or more flexible pools of logical storage devices that may be virtualized to maintain a server's virtual storage device). In various embodiments, server 120 may run one or more services or software applications that provide the functionality described below.

[0033] The computing units in the server 120 may run one or more operating systems including any of the above operating systems and any commercially available server operating systems. The server 120 may also run any of a variety of additional server applications and / or middle-tier applications, including HTTP servers, FTP servers, CGI servers, JAVA servers, database servers, etc.

[0034] In some implementations, server 120 may include one or more applications to analyze and consolidate data feeds and / or event updates received from users of client devices 101, 102, 103, 104, 105, and / or 106. Server 120 may also include one or more applications to display data feeds and / or real-time events via one or more display devices of client devices 101, 102, 103, 104, 105, and / or 106.

[0035] In some embodiments, the server 120 may be a server of a distributed system, or a server combined with a blockchain. The server 120 may also be a cloud server, or an intelligent cloud computing server or intelligent cloud host with artificial intelligence technology. A cloud server is a host product in a cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and virtual private servers (VPS) services.

[0036] The system 100 may also include one or more databases 130. In some embodiments, these databases may be used to store data and other information. For example, one or more of the databases 130 may be used to store information such as audio files and video files. The databases 130 may reside in various locations. For example, the database used by the server 120 may be local to the server 120, or may be remote from the server 120 and may communicate with the server 120 via a network-based or dedicated connection. The databases 130 may be of different types. In some embodiments, the databases used by the server 120 may be, for example, relational databases. One or more of these databases may store, update, and retrieve data to and from the databases in response to commands.

[0037] In some embodiments, one or more of the databases 130 may also be used by applications to store application data. The databases used by the applications may be different types of databases, such as a key-value store, an object store, or a conventional store backed by a file system.

[0038] Figure 1The system 100 may be configured and operated in various ways to enable the application of various methods and apparatuses described in the present disclosure.

[0039] Various aspects of the artificial intelligence-based content generation method according to an embodiment of the present disclosure will be described in detail below with reference to the accompanying drawings.

[0040] According to some embodiments, the artificial intelligence in the present disclosure may be a large language model (also referred to as a generative language model or a deep learning large model). A large language model is a deep learning model for natural language processing based on a neural network, and its goal is to generate natural language text that conforms to language rules based on a given input. The large language model captures vocabulary, grammar, semantics, and contextual information by learning a large corpus, thereby generating coherent and fluent sentences or paragraphs when given an input or context. Although the current large language models mostly use the Transformer architecture, it should be pointed out that the artificial intelligence of the present disclosure is not limited to large language models using the Transformer architecture, but also includes other artificial intelligence models, architectures, or systems that use other architectures and in which the present invention can be combined.

[0041] Large language models have end-to-end characteristics and can directly generate response data based on user input data without the help of functional components or other inputs outside the large language model. In other words, the large language model itself has a generative function. Large language models usually have billions or even hundreds of billions of parameters, and they are usually trained on large-scale text data or other modal data. Large language models can be used for various natural language processing tasks, such as text generation, language translation, and question-answering systems.

[0042] Figure 2 A flow chart of a method 200 for generating content based on artificial intelligence according to an embodiment of the present disclosure is shown. In some embodiments, the method 200 may be performed using a content generation system (eg, the system 100 described above).

[0043] like Figure 2 As shown, method 200 includes step S201, step S202, step S203 and step S204.

[0044] In step S201, the user's query information is obtained.

[0045] The query information may include information in text, images, audio, video and / or other modalities, and may be in any form. The query information may be provided directly by the user, or may be obtained by processing the original query content provided by the user, as will be described below. In some embodiments, the content generation system may provide an interface for interacting with the user, and the user may enter a question through a text box, voice input, or submit query information or original query content through other interactive methods, so that the content generation system can obtain the user's query information or original query content based on the interface.

[0046] In step S202, the query information is structurally understood to obtain a content generation strategy corresponding to the query information, wherein the content generation strategy includes two or more sub-questions associated with the query information.

[0047] In some embodiments, structured understanding may include parsing the user's query information and converting it into a structured data form. Natural Language Processing (NLP) techniques, such as semantic analysis, entity recognition, intent recognition, etc., may be used to understand the user's query intent and / or context. Based on the results of structured understanding, artificial intelligence will formulate a content generation strategy. This strategy is a framework or plan to guide content generation, which will determine how to respond to the user's query. Accordingly, the content generation strategy not only directly responds to the user's query, but can further generate two or more sub-questions associated with the user's query.

[0048] In an exemplary embodiment, the user's query information is "aging research", and the content generation strategy obtained through structured understanding can be "the current situation of aging", "the causes and development trends of aging", "the impact of aging", "the impact of aging on social economy", "the impact of aging on different groups of people" and "coping strategies for aging".

[0049] In step S203, a user interaction interface is provided to output the content generation strategy and receive adjustments made by the user to the content generation strategy.

[0050] In some embodiments, the user interaction interface is an interface for communication and interaction between a person and a computer system. It allows the user to communicate with the system in some way, input instructions or data, and receive feedback from the system. The user interaction interface may include a graphical user interface, a command line interface, a voice recognition interface, a touch screen interface, a gesture control interface, a chat interface, and other interfaces that can realize human-computer communication and interaction. The user can view the generated content generation strategy through the user interaction interface, and propose amendments or adjustments to the generated content generation strategy according to his or her needs. In an exemplary embodiment, the above-mentioned adjustments may include changing sub-questions, adding or deleting information points, and / or adjusting the style or tone of the reply.

[0051] After receiving the user's adjustment opinion, the content generation system updates the content generation strategy and generates reply content according to the updated content generation strategy. In some embodiments, the update process may include modifying a preset template, redefining key information points, adjusting algorithm parameters, changing the retrieval method and / or retrieval database. Furthermore, for the generated reply content corresponding to the query information, the content generation strategy may be adjusted again via the user interaction interface to further improve the reply content.

[0052] In step S204, based on the adjusted content generation strategy, reply content corresponding to the query information is generated.

[0053] In some embodiments, the adjusted content generation strategy and query information may be input into a pre-trained large language model to obtain reply content generated by the large language model.

[0054] In the content generation method 200 based on artificial intelligence according to the embodiment of the present disclosure, a new paradigm is proposed for artificial intelligence and users to collaboratively generate high-quality content. Specifically, artificial intelligence first formulates a content generation strategy based on a structured understanding of the query information, and then the user participates in and adjusts the strategy to obtain an adjusted content generation strategy, and finally uses the content generation strategy collaboratively determined by artificial intelligence and users to generate personalized reply content. In this way, it can be ensured that the generated reply content is more comprehensive and in-depth, covering all aspects that the user may be concerned about, and at the same time, it can also improve the efficiency and quality of reply content generation, achieving a more accurate and personalized effect.

[0055] Figure 3 The flowchart of the process 300 of acquiring the query information of the user according to the embodiment of the present disclosure is shown. The process 300 can be used to implement the above step S201.

[0056] like Figure 3 As shown, process 300 includes step S301, step S302 and step S303.

[0057] In step S301, an original query content from a user is received.

[0058] In some embodiments, the original query content of the user may be text, voice, image or other forms of data. The original query content may be a question, a request or any information that requires a response.

[0059] In step S302, the original query content is input into a pre-trained first language model to obtain the user's intention, which represents the text category of the reply content to be generated.

[0060] In some embodiments, the original query information can be input into a pre-trained first language model. The pre-trained first language model can be a model based on deep learning and trained by supervised fine-tuning (SFT), such as ERNIE, BERT, GPT or other natural language processing models, which has been trained to have the ability to understand a large amount of text data and language patterns.

[0061] The user's intent may indicate a specific goal or need that the user hopes to achieve through the query. In an exemplary embodiment, the user's intent may include obtaining information, solving problems, and / or performing specific tasks. The pre-trained first language model uses its learned language knowledge and pattern recognition capabilities to infer the user's intent by analyzing the input query content. This intent may be related to the text category of the reply content to be generated. Exemplary text categories may include: copywriting text, question and answer text, code text, and / or consumer decision text. In an exemplary embodiment, the user's original query content is "the current status of aging", and the user's intent obtained from the first language model may be to generate a copywriting text reply content.

[0062] In some embodiments, the first language model may be trained based on a neural network of a Transformer architecture, the training sample input value may include the user's original query content, and the output value may include a reference output (groundtruth) of the intent.

[0063] In step S303, the user's query information is generated based on the original query content and intent.

[0064] In some embodiments, the user's original query content and the acquired user's intention can be combined in various ways to generate the user's query information. Based on the original query content and the user's true intention, artificial intelligence can be helped to generate query information that is easier to understand and recall relevant knowledge.

[0065] In the process 300 of obtaining the user's query information according to the embodiment of the present disclosure, the user's intent is obtained through the trained first language model, and the query information is generated based on the original query content and the intent, which can ensure that the generated query information matches the user's intent and needs, thereby improving the quality of the generated content and providing a better user experience.

[0066] In some embodiments, Figure 3 The step S303 shown, generating the user's query information based on the original query content and intent, may include: obtaining a first search result associated with the original query content; and generating the user's query information based on the first search result, the original query content and intent.

[0067] In some embodiments, the first search result associated with the original query content may include the user's historical conversation information, the user's personalized information, the user's search information, reference information related to the user's original query content, and / or contextual rewriting of the original query content. The user's original query content, the acquired user's intent, and the above-mentioned first search result may be combined in various ways to generate the user's query information.

[0068] Therefore, based on the user's original query and intention, combined with the first search results obtained externally, the query information can be supplemented, which is conducive to improving the quality of generated query information and further improving the quality of generated reply content.

[0069] In some embodiments, the first part of the search results for the original query content may be a direct response to the query or question originally raised by the user. In an exemplary embodiment, the user query is "how to make pasta", and the first part of the search results may include directly related information such as recipes, steps, and techniques for making pasta. The search results may include text, pictures, videos, etc. according to the user's needs.

[0070] In some embodiments, when the user's query or question contains multiple sub-questions or requires more detailed information, the second part of the search results may include more specific information for these sub-questions. In an exemplary embodiment, the user query is "how to make pasta", the first part of the search results may show the general method of making pasta, while the second part of the search results may provide specific methods and recipes for different types of pasta (such as spaghetti, macaroni, etc.), and may also include user comments, expert opinions, related forum discussions, etc. The search results may include text, pictures, audio, video and other forms according to the user's needs.

[0071] In some embodiments, step S202, performing structured understanding on the query information to obtain a content generation strategy corresponding to the query information may include: inputting the query information into a pre-trained second language model different from the first language model to obtain the content generation strategy.

[0072] In some embodiments, a user may submit a query through a client device, such as "How can I improve my English speaking ability?". The first language model may be a pre-trained general large language model, and the content generation strategy obtained through structured understanding may include key words and phrases, such as "improve", "English speaking", and "ability". The second language model may be a pre-trained professional large language model, which is specifically used to generate content generation strategies related to the query information and can more deeply understand the specific context and needs of a specific field. For example, the second language model may be a large language model in the field of education, and the content generation strategy obtained through structured understanding may be "What are the basic elements of spoken English?", "What are the scientifically proven English speaking practice techniques?", "How to integrate English speaking practice into daily activities?".

[0073] Therefore, through the pre-trained large language model, the user's intention can be understood more accurately and a content generation strategy that better meets the user's needs can be generated. The specialized training of the second language model for a specific field makes it more accurate in processing queries in a specific field, improving the quality and efficiency of content generation and further meeting the personalized needs of users.

[0074] In some embodiments, the second language model can be trained based on a neural network of a Transformer architecture, the training sample input value can include the user's query information, and the output value can include the reference output (ground truth) of the content generation strategy. In some embodiments, the input value can also include external knowledge associated with the query information.

[0075] In some embodiments, the user interaction interface provided in step S203 may be a graphical user interface (GUI).

[0076] In an exemplary embodiment, a user submits a query through a text input box of the GUI: "How should I prepare for an upcoming marathon?", and the system backend will generate related sub-questions: "marathon training plan", "nutritional intake recommendations", and "equipment selection".

[0077] In some embodiments, the graphical user interface can display a strategy list, each sub-question can be equipped with a check box, and buttons such as "Edit", "Select", and "Add" can be set for users to select and modify. The user can select the "Marathon training plan" and "Equipment selection" sub-questions through the GUI, and can click the "Edit" button to modify the "Nutrition intake recommendation" sub-question to "Pre-race nutrition and diet plan".

[0078] Therefore, the use of graphical user interfaces greatly improves the convenience of user interaction with the system. Users can intuitively see the content generation strategy and easily select, edit, and add new operations. This direct interaction method improves the personalization of content generation and improves the user experience.

[0079] In some embodiments, the adjustment made by the user to the content generation strategy may include at least one of the following: deleting one or more sub-questions of the two or more sub-questions; modifying one or more sub-questions of the two or more sub-questions; and adding one or more sub-questions.

[0080] In an exemplary embodiment, the query content submitted by the user, such as "How can I improve my sleep quality?", is processed by a pre-trained language model, and an initial content generation strategy is proposed, including the following sub-questions: "Methods to improve the sleeping environment", "The impact of eating habits on sleep", "Pre-bedtime relaxation techniques", and "Identification and treatment of sleep disorders".

[0081] In some embodiments, users can use tools such as "Delete", "Edit" and "Add" to adjust the content generation strategy according to their needs and preferences. In an exemplary embodiment, when the user believes that "Pre-sleep relaxation techniques" and "Identification and treatment of sleep disorders" are not very relevant to their needs, they can click the "Delete" button for these two sub-questions to remove them from the content generation strategy. When the user wants to know more specifically about the relationship between eating habits and improving sleep, they can click the "Edit" button next to "The impact of eating habits on sleep" and modify it to "Foods and eating habits that help improve sleep." If the user wants to add information about pre-sleep exercise, he can click the "Add" button to enter a new sub-question "Appropriate amount of exercise before bedtime."

[0082] Therefore, by providing flexible adjustment methods, users can accurately express their expectations for generated content, improving the relevance and personalization of content to cover more specific needs of users.

[0083] In some embodiments, the artificial intelligence-based content generation method may also include: after obtaining the content generation strategy corresponding to the query information, at least inputting the content generation strategy, the query information, and external knowledge associated with the query information into a pre-trained third language model to obtain an evaluation result characterizing whether the content generation strategy meets the preset strategy standards.

[0084] In an exemplary embodiment, a user submits a query "How to prevent common office occupational diseases?" through the content generation system, and the system processes the user query through the pre-trained first language model and the second language model to generate a content generation strategy. The content generation strategy may include the following sub-questions: "Types of office occupational diseases", "Preventive measures for office occupational diseases", and "Healthy work habits". The external knowledge collected by the system related to the user query may include: the latest research medical journal papers, guidelines issued by the government or the World Health Organization (WHO), and blog posts by well-known health experts. The user's query information, the sub-questions in the content generation strategy, and the collected external knowledge can be integrated into an input data set. For example, "Types of office occupational diseases" can correspond to "The latest academic research shows that long-term sitting work may cause low back pain"; "Preventive measures for occupational diseases" can correspond to "WHO guidelines recommend regular standing and stretching to reduce the risk of occupational diseases"; "Healthy work habits" can correspond to "Health expert blogs recommend 5 minutes of activity every hour."

[0085] In some embodiments, the integrated input data set can be input into a pre-trained third language model to evaluate the quality of the content generation strategy, which can focus on the relevance, completeness, practicality, timeliness, and / or authority of each sub-question. Relevance can be used to evaluate whether the sub-question directly responds to the core needs of the user's query; completeness can be used to check whether all important aspects are covered and no key information is omitted; practicality can be used to evaluate whether the information provided is actionable and whether the user can directly apply it; timeliness can be used to evaluate whether the information is up to date by comparing the latest research with old data; authority can be used to identify the source of information (e.g., government health departments or well-known academic institutions) to evaluate the reliability of the information.

[0086] In an exemplary embodiment, for "types of office occupational diseases", the third language model can be evaluated as highly relevant and authoritative through the detailed classification provided by the latest research papers. For "preventive measures for occupational diseases", the latest guidelines issued by WHO can be identified and evaluated as highly timely and authoritative. For "healthy work habits", practical advice from blog posts by health experts can be provided, and the qualifications of the experts and the publication time of the articles can be further verified to evaluate them as highly timely and authoritative. Through the evaluation results, the advantages and disadvantages of each sub-question can be pointed out, and the recommendation of the overall content generation strategy can be given.

[0087] Therefore, by introducing the evaluation process of the pre-trained third language model, after using artificial intelligence to generate a preliminary content generation strategy, the artificial intelligence can first evaluate whether the preliminary strategy meets the preset strategy standards based on query information and external knowledge, and then the user can adjust it. This method can ensure that the content generation strategy has undergone a round of internal verification and review before being provided to the user, thereby improving the quality of the content generation strategy, reducing the burden of subsequent adjustments and modifications on the user, and improving the user experience. In addition, this evaluation method ensures that the content generation strategy obtained after the evaluation not only meets the basic needs of users, but also improves the accuracy and reliability of the content, and provides the latest and reliable information.

[0088] In some embodiments, the evaluation result may include a classification result of whether the content generation strategy meets the preset policy standards. In response to the third language model determining that the input content generation strategy does not meet the preset policy standards, the evaluation result generated by it may further include a regenerated content generation strategy. In other words, the third language model determines that the current content generation strategy does not meet the preset policy standards based on the input content, and it can output the conclusion (i.e., the content generation strategy does not meet the preset policy standards) and directly output the regenerated content generation strategy. In an exemplary embodiment, in response to the third language model determining that the input content generation strategy does not meet the preset policy standards, the evaluation result may further include a plurality of regenerated sub-questions.

[0089] In some embodiments, after obtaining the content generation strategy corresponding to the query information, a summary of the content generation strategy can be generated based on the query information and the external knowledge associated with the query information. The generated summary can be input into a pre-trained third language model together with the content generation strategy, the query information, and the external knowledge associated with the query information to obtain a more accurate evaluation result.

[0090] In some embodiments, the third language model can be trained based on a neural network of a Transformer architecture, the training sample input value can include the user's content generation strategy, query information, and external knowledge associated with the query information, and the output value can include a reference output (ground truth) of an evaluation result representing whether the content generation strategy meets the preset strategy standard. In some embodiments, the output value can also include a reference output of the adjusted content generation strategy.

[0091] According to some embodiments, the evaluation process of the content generation strategy and the process of re-structuring the understanding can be decoupled. The content generation method based on artificial intelligence can also include: storing at least the content generation strategy, the query information, the external knowledge and the evaluation result in a memory; and in response to judging that the evaluation result does not meet the preset strategy standard, retrieving the content generation strategy, the query information, the external knowledge and the evaluation result from the memory, and re-structuring the understanding.

[0092] In some embodiments, after obtaining the content generation strategy, the relevant information of this structured understanding (including but not limited to the currently obtained content generation strategy, query information, external knowledge and evaluation results) can be stored in a memory bank (i.e., a memory). In response to judging that the evaluation result does not meet the preset strategy standard, the structured understanding can be re-performed in combination with the existing information in the memory bank (e.g., at least one previously obtained content generation strategy and the relevant information of the previous structured understanding). In some embodiments, the content generation strategy, query information, external knowledge and evaluation results retrieved from the memory can be input into the second language model to obtain the regenerated content generation strategy.

[0093] Therefore, the above method can record and utilize the relevant information of each structured understanding, and effectively utilize this information when regenerating the content generation strategy to improve the quality of the content generation strategy and make it more in line with user needs and expectations.

[0094] In some embodiments, the adjusted content generation strategy may include two or more adjusted sub-questions. Step S204, based on the adjusted content generation strategy, generating reply content corresponding to the query information may include: obtaining a second retrieval result associated with the adjusted two or more sub-questions; and generating the reply content based on the second retrieval result and the query information.

[0095] Therefore, after the user adjusts the content generation strategy, by searching the sub-questions included therein, it is possible to generate reply content using the search results related to the strategy adjusted by the user, thereby improving the quality of the reply content.

[0096] In some embodiments, in addition to the second search result and the query information, the response content may be generated using an adjusted content generation strategy, the original query content and / or the user's intent as a reference to further improve the fit between the response content and the user's needs.

[0097] In some embodiments, a pre-trained large language model may be used to generate reply content in step S204. The pre-trained large language model may be trained based on a neural network of a Transformer architecture, and the input value of the training sample may include the user's query information and the search results associated with the content generation strategy, and the output value may include the reference output (ground truth) of the reply content. In some embodiments, the input value of the training sample may also include the content generation strategy, the original query content and / or the user's intention.

[0098] According to some embodiments, the artificial intelligence-based content generation method may also include: providing users with the function of subscribing to topics related to the query information; and in response to the user subscribing to the topic and in response to monitoring updated information related to the topic, pushing reply content regenerated based on the updated information to the user.

[0099] In some embodiments, when a user wants to subscribe to a topic related to the current query, the subscription can be achieved by clicking a subscription button or other methods. After completing the subscription, the content generation system can detect updated information related to the topic, and after detecting the update, regenerate the reply content and push it to the user.

[0100] Therefore, through the above method, after important information appears on the topic that the user cares about, new reply content can be generated in time and pushed to the user, so that the user can grasp the valuable information in time.

[0101] In some embodiments, considering that hot topics may update information frequently, and the sources and quality of information vary, the content generation system may determine whether the updated information is more valuable or whether it is necessary to regenerate an answer and push it to the user after detecting an update. An evaluation standard for whether it is more valuable may be formulated a priori, and when it is determined that the updated information meets the evaluation standard, the answer is regenerated and pushed to the user.

[0102] According to some embodiments, in response to the user subscribing to the topic, and in response to monitoring updated information related to the topic, pushing reply content regenerated based on the updated information to the user may include: inputting the query information, the updated information, and the historically generated reply content into a pre-trained fourth language model to obtain an evaluation result characterizing whether the updated information meets a preset value standard.

[0103] In an exemplary embodiment, the preset value criteria may include relevance, completeness, practicality, timeliness and / or authority of the information. These evaluation criteria may refer to the above description and are not described in detail here.

[0104] Therefore, by using the fourth language model to combine the original query information and the historically generated response content to complete the value assessment of the updated information, more accurate evaluation results can be obtained, thereby improving the user experience.

[0105] In some embodiments, in response to the fourth language model determining that the input updated information meets the preset value standard, the evaluation result generated by it may further include the regenerated reply content. In other words, the fourth language model determines that the updated information meets the evaluation standard based on the input content, and it can output the conclusion (that is, the updated information has value) and directly output the regenerated reply content.

[0106] Therefore, by directly regenerating the reply content when it is determined that the updated information has value, the regenerated reply content can be easily acquired.

[0107] In some embodiments, the fourth language model can be trained based on a neural network of a Transformer architecture, the training sample input value can include query information, updated information, and historically generated reply content, and the output value can include a reference output (ground truth) of the evaluation result. In some embodiments, the output value of the training sample can also include a reference output (ground truth) of the adjusted reply content.

[0108] In some embodiments, the value assessment process of the updated information and the process of regenerating the reply content may also be decoupled. In response to determining that the assessment result output by the fourth language model indicates that the updated information meets the preset value standard, the updated information may be used as a retrieval result associated with the user's original query content to regenerate updated query information, and then the reply content corresponding to the updated query information is generated using the pre-trained large language model and pushed to the user.

[0109] Figure 4A schematic diagram of an artificial intelligence-based content generation architecture 400 according to an embodiment of the present disclosure is shown. In some embodiments, an original query content 402 from a user may be first received, and the original query content may be input into a pre-trained first language model to obtain the user's intent 404. At the same time, a first search result 406 associated with the original query content 402 may be obtained. After obtaining the intent 404 and the first search result 406, the user's query information 408 may be generated based on the two (and optionally, the original query content 402). Furthermore, the query information 408 may be structured and understood using a second language model to obtain a content generation strategy 410 corresponding to the query information 408. A third language model may be used to evaluate whether the content production strategy meets the preset strategy criteria 412, and a content generation strategy 414 may be regenerated when it is determined that it does not meet the criteria. A user interaction interface may be provided to receive a user's adjustment 416 to the content generation strategy. Then, a pre-trained large language model may be used to generate reply content 418 based on the adjusted content generation strategy. In addition, a topic subscription function may be provided, and when updated information related to the topic is detected, the reply content is regenerated and pushed to the user to implement subscription push 420.

[0110] The embodiments of the present disclosure also provide a content generation device based on artificial intelligence.

[0111] like Figure 5 A structural block diagram of an artificial intelligence-based content generation device 500 according to an embodiment of the present disclosure is shown.

[0112] like Figure 5 As shown, the device 500 includes an information acquisition module 501, a strategy generation module 502, an interaction module 503 and a reply module 504. The information acquisition module 501 is configured to obtain the user's query information. The strategy generation module 502 is configured to perform a structured understanding of the query information to obtain a content generation strategy corresponding to the query information, wherein the content generation strategy includes two or more sub-questions associated with the query information. The interaction module 503 is configured to provide a user interaction interface for outputting the content generation strategy and receiving adjustments made by the user to the content generation strategy. The reply module 504 is configured to generate reply content corresponding to the query information based on the adjusted content generation strategy.

[0113] The operations of the information acquisition module 501, the strategy generation module 502, the interaction module 503 and the reply module 504 may correspond to the following: Figure 2 The steps S201, S202, S203 and S204 are shown in FIG. Therefore, the details of each aspect are not repeated here.

[0114] Figure 6A structural block diagram of an artificial intelligence-based content generation device 600 according to another embodiment of the present disclosure is shown.

[0115] like Figure 6 As shown, the apparatus 600 may include an information acquisition module 601, a strategy generation module 602, an interaction module 603, and a reply module 604. The operations of the above modules may be similar to those of Figure 5 The operations of the information acquisition module 501, the strategy generation module 502, the interaction module 503 and the reply module 504 are the same. In addition, the above modules may also include further sub-modules.

[0116] In some embodiments, the information acquisition module 601 may include: a receiving module 6011, configured to receive original query content from a user; an intention acquisition module 6012, configured to input the original query content into a pre-trained first language model to obtain the user's intention, wherein the intention represents the text category of the reply content to be generated; and an information generation module 6013, configured to generate the user's query information based on the original query content and the intention.

[0117] In some embodiments, the information generation module 6013 may include: a first result acquisition module 6013a, configured to acquire a first retrieval result associated with the original query content; and an information generation submodule 6013b, configured to generate the user's query information based on the first retrieval result, the original query content and the intent.

[0118] In some embodiments, the adjusted content generation strategy may include two or more adjusted sub-questions. The reply module 604 may include: a second result acquisition module 6041, configured to acquire a second search result associated with the two or more adjusted sub-questions; and a reply generation submodule 6042, configured to generate the reply content based on the second search result and the query information.

[0119] In some embodiments, the strategy generation module 602 may include: a strategy generation submodule 6021 configured to input query information into a pre-trained second language model different from the first language model to obtain a content generation strategy.

[0120] In some embodiments, the user interaction interface may be a graphical user interface.

[0121] In some embodiments, the reply module 604 may include at least one of the following (not shown in the figure): a deletion module, configured to delete one or more sub-questions among two or more sub-questions; a modification module, configured to modify one or more sub-questions among two or more sub-questions; and an addition module, configured to add one or more sub-questions.

[0122] In some embodiments, the device 600 may also include: an evaluation module 605, which is configured to, after obtaining the content generation strategy corresponding to the query information, at least input the content generation strategy, the query information, and external knowledge associated with the query information into a pre-trained third language model to obtain an evaluation result characterizing whether the content generation strategy meets the preset strategy standards.

[0123] In some embodiments, the device 600 may also include: a storage module 606, configured to store at least the content generation strategy, the query information, the external knowledge and the evaluation result in a memory; and a repetition module 607, configured to, in response to a judgment that the evaluation result does not meet the preset strategy standard, retrieve the content generation strategy, the query information, the external knowledge and the evaluation result from the memory and re-structure the understanding.

[0124] In some embodiments, the device 600 may also include: a subscription module 608, configured to provide the user with a function of subscribing to topics related to the query information; and a push module 609, configured to push reply content regenerated based on the updated information to the user in response to the user subscribing to the topic and in response to monitoring updated information related to the topic.

[0125] In some embodiments, the push module 609 may be configured to input the query information, the updated information, and the historically generated reply content into a pre-trained fourth language model to obtain an evaluation result indicating whether the updated information meets a preset value standard.

[0126] In the technical solution of the present disclosure, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0127] According to an embodiment of the present disclosure, an electronic device is also provided, comprising at least one processor; and a memory communicatively connected to the at least one processor, wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method as described above.

[0128] According to an embodiment of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is further provided, wherein the computer instructions are used to enable a computer to execute the method as described above.

[0129] According to an embodiment of the present disclosure, a computer program product is also provided, including a computer program, wherein the computer program implements the method as described above when executed by a processor.

[0130] refer to Figure 7 , a block diagram of an electronic device 700 that can be used as a server or client of the present disclosure will now be described, which is an example of a hardware device that can be applied to various aspects of the present disclosure. The electronic device is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0131] like Figure 7 As shown, the electronic device 700 includes a computing unit 701, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 702 or a computer program loaded from a storage unit 708 into a random access memory (RAM) 703. In the RAM 703, various programs and data required for the operation of the electronic device 700 can also be stored. The computing unit 701, the ROM 702, and the RAM 703 are connected to each other via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0132] Multiple components in the electronic device 700 are connected to the I / O interface 705, including: an input unit 706, an output unit 707, a storage unit 708, and a communication unit 709. The input unit 706 can be any type of device that can input information to the electronic device 700. The input unit 706 can receive input digital or character information and generate key signal input related to user settings and / or function control of the electronic device, and can include but is not limited to a mouse, a keyboard, a touch screen, a track pad, a track ball, a joystick, a microphone, and / or a remote controller. The output unit 707 can be any type of device that can present information, and can include but is not limited to a display, a speaker, a video / audio output terminal, a vibrator, and / or a printer. The storage unit 708 can include but is not limited to a disk, an optical disk. The communication unit 709 allows the electronic device 700 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks, and can include but is not limited to a modem, a network card, an infrared communication device, a wireless communication transceiver, and / or a chipset, such as a Bluetooth device, an 802.11 device, a WiFi device, a WiMax device, a cellular communication device, and / or the like.

[0133] The computing unit 701 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 701 performs the various methods and processes described above. For example, in some embodiments, the method may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 708. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 700 via the ROM 702 and / or the communication unit 709. When the computer program is loaded into the RAM 703 and executed by the computing unit 701, one or more steps of the method described above may be performed. Alternatively, in other embodiments, the computing unit 701 may be configured to perform the method described above in any other appropriate manner (e.g., by means of firmware).

[0134] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0135] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0136] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0137] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0138] The systems and techniques described herein can be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), the Internet, and a blockchain network.

[0139] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0140] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0141] Although the embodiments or examples of the present disclosure have been described with reference to the accompanying drawings, it should be understood that the above-mentioned methods, systems and devices are merely exemplary embodiments or examples, and the scope of the present invention is not limited by these embodiments or examples, but only by the claims after authorization and their equivalent scope. Various elements in the embodiments or examples can be omitted or replaced by their equivalent elements. In addition, each step can be performed in an order different from that described in the present disclosure. Further, the various elements in the embodiments or examples can be combined in various ways. It is important that with the evolution of technology, many elements described herein can be replaced by equivalent elements that appear after the present disclosure.

Claims

1. A content generation method based on artificial intelligence, the method comprising: Get the user's query information; Performing a structured understanding on the query information to obtain a content generation strategy corresponding to the query information, wherein the content generation strategy includes two or more sub-questions associated with the query information; Providing a user interaction interface for outputting the content generation strategy and receiving adjustments made by the user to the content generation strategy; and Based on the adjusted content generation strategy, reply content corresponding to the query information is generated.

2. The method according to claim 1, wherein: The obtaining of the user's query information includes: Receiving original query content from the user; Inputting the original query content into a pre-trained first language model to obtain the user's intention, wherein the intention represents the text category of the reply content to be generated; and The user's query information is generated based on the original query content and the intention.

3. The method according to claim 2, wherein: The generating the user's query information based on the original query content and the intention includes: Obtaining a first search result associated with the original query content; and The user's query information is generated based on the first search result, the original query content and the intention.

4. The method according to any one of claims 1 to 3, wherein: The adjusted content generation strategy includes two or more adjusted sub-questions, and generating reply content corresponding to the query information based on the adjusted content generation strategy includes: Obtaining second search results associated with the adjusted two or more sub-questions; and The reply content is generated based on the second search result and the query information.

5. The method according to any one of claims 1 to 4, wherein: The structurally understanding the query information to obtain a content generation strategy corresponding to the query information includes: The query information is input into a pre-trained second language model different from the first language model to obtain the content generation strategy.

6. The method according to any one of claims 1 to 5, wherein: The user interaction interface is a graphical user interface.

7. The method according to any one of claims 1 to 6, wherein: The adjustment made by the user to the content generation strategy includes at least one of the following: deleting one or more sub-questions among the two or more sub-questions; modifying one or more of the two or more sub-questions; and Add one or more sub-questions.

8. The method according to any one of claims 2 to 7, wherein: The method further comprises: After obtaining the content generation strategy corresponding to the query information, at least the content generation strategy, the query information, and external knowledge associated with the query information are input into a pre-trained third language model to obtain an evaluation result characterizing whether the content generation strategy meets a preset strategy standard.

9. The method according to claim 8, wherein: The method further comprises: storing at least the content generation strategy, the query information, the external knowledge, and the evaluation result in a memory; and In response to determining that the evaluation result does not meet the preset strategy standard, the content generation strategy, the query information, the external knowledge and the evaluation result are retrieved from the memory and restructured and understood.

10. The method according to any one of claims 1 to 9, wherein: The method further comprises: Providing the user with a function of subscribing to topics related to the query information; and In response to the user subscribing to the topic and in response to monitoring updated information related to the topic, a reply content regenerated based on the updated information is pushed to the user.

11. The method according to claim 10, wherein: In response to the user subscribing to the topic and in response to monitoring updated information related to the topic, pushing to the user a reply content regenerated based on the updated information includes: The query information, the updated information and the historically generated reply content are input into a pre-trained fourth language model to obtain an evaluation result indicating whether the updated information meets a preset value standard.

12. A content generation device based on artificial intelligence, the device comprising: An information acquisition module is configured to acquire query information of a user; a strategy generation module, configured to perform a structured understanding on the query information to obtain a content generation strategy corresponding to the query information, wherein the content generation strategy includes two or more sub-questions associated with the query information; an interactive module configured to provide a user interactive interface for outputting the content generation strategy and receiving adjustments made by the user to the content generation strategy; and The reply module is configured to generate reply content corresponding to the query information based on the adjusted content generation strategy.

13. The device according to claim 12, wherein: The information acquisition module includes: A receiving module is configured to receive original query content from a user; an intention acquisition module configured to input the original query content into a pre-trained first language model to acquire the user's intention, wherein the intention represents a text category of the reply content to be generated; and The information generation module is configured to generate the user's query information based on the original query content and the intention.

14. The device according to claim 13, wherein: The information generation module comprises: A first result acquisition module is configured to acquire a first search result associated with the original query content; and The information generation submodule is configured to generate the user's query information based on the first search result, the original query content and the intention.

15. The device according to claim 14, wherein: The adjusted content generation strategy includes two or more adjusted sub-questions, and the response module includes: a second result acquisition module configured to acquire a second search result associated with the two or more adjusted sub-questions; and The reply generation submodule is configured to generate the reply content based on the second search result and the query information.

16. The device according to any one of claims 12 to 14, wherein: The strategy generation module comprises: The strategy generation submodule is configured to input the query information into a pre-trained second language model different from the first language model to obtain the content generation strategy.

17. The device according to any one of claims 12 to 16, wherein: The user interaction interface is a graphical user interface.

18. The device according to any one of claims 12 to 17, wherein: The reply module includes at least one of the following: a deletion module configured to delete one or more sub-questions among the two or more sub-questions; a modification module configured to modify one or more sub-questions of the two or more sub-questions; and The adding module is configured to add one or more sub-questions.

19. The device according to any one of claims 13 to 18, wherein: The device also includes: The evaluation module is configured to, after obtaining the content generation strategy corresponding to the query information, at least input the content generation strategy, the query information, and external knowledge associated with the query information into a pre-trained third language model to obtain an evaluation result characterizing whether the content generation strategy meets the preset strategy standard.

20. The device according to claim 19, wherein The device also includes: a storage module configured to store at least the content generation strategy, the query information, the external knowledge, and the evaluation result in a memory; and The repetition module is configured to retrieve the content generation strategy, the query information, the external knowledge and the evaluation result from the memory in response to determining that the evaluation result does not meet the preset strategy standard, and re-structure the understanding.

21. The device according to any one of claims 12 to 20, wherein: The device also includes: a subscription module configured to provide the user with a function of subscribing to topics related to the query information; and The push module is configured to push the reply content regenerated based on the updated information to the user in response to the user subscribing to the topic and in response to monitoring updated information related to the topic.

22. The device according to claim 21, wherein The push module is configured to input the query information, the updated information and the historically generated reply content into a pre-trained fourth language model to obtain an evaluation result representing whether the updated information meets a preset value standard.

23. An electronic device, comprising: at least one processor; as well as a memory communicatively coupled to the at least one processor; in The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 11.

24. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-11.

25. A computer program product comprising a computer program, wherein: When the computer program is executed by a processor, the method according to any one of claims 1 to 11 is implemented.

Images