Text generation method, text generation system, computing device, computer readable storage medium and computer program product

Abstract

The embodiment of the invention provides a text generation method, a text generation system, computing equipment, a computer readable storage medium and a computer program product. The method comprises the steps that a text generation task and task execution information corresponding to the text generation task are determined; determining a target execution sub-process in a preset execution process according to the task execution information; selecting a function interface corresponding to the text generation task in a preset interface set corresponding to the target field according to the target execution sub-process; and calling the function interface to execute the text generation task according to the target execution sub-process to obtain a target text corresponding to the text generation task. The intelligent decision component automatically decides to complete the execution process of the text generation task, so that the labor cost is reduced, the text generation efficiency is improved, the function interface corresponding to the text generation task is selected according to the target execution sub-process, the function interface is called to complete the text generation task, the cost overhead caused by manual processing is reduced, and the text generation efficiency is improved. And the text generation complexity is reduced.

Description

Technical Field

[0001] The embodiments of this specification relate to the field of artificial intelligence technology, and in particular to a text generation method, a text generation system, a computing device, a computer-readable storage medium, and a computer program product. Background Art

[0002] With the development of artificial intelligence technology, natural language processing has been widely used in various fields, such as intelligent question and answer, customer service consultation, online teaching, online shopping, etc. In the field of text creation, such as news release creation and magazine article writing, very professional knowledge is required, and there are very high requirements for the content of the creation. At present, in the manual creation process, it is often necessary to manually collect and organize information, write content layout and other operations. After the creation is completed, the creative results need to be proofread, reviewed, refined and other steps before the article can be finally completed. This is inefficient and labor-intensive. Therefore, how to automatically create high-quality articles that meet the user's creative needs while reducing labor costs and improving processing efficiency is a problem that needs to be solved urgently. Summary of the Invention

[0003] In light of this, embodiments of this specification provide a text generation method. One or more embodiments of this specification also involve a text generation apparatus, a text generation system, a computing device, a computer-readable storage medium, and a computer program product to address the existing technology's inability to efficiently and automatically create high-quality articles in professional fields.

[0004] According to a first aspect of an embodiment of this specification, a text generation method is provided, which is applied to an intelligent decision-making component, comprising:

[0005] Determining a text generation task and task execution information corresponding to the text generation task, wherein the task execution information is used to determine task requirements of the text generation task;

[0006] Determining a target execution sub-process in a preset execution process based on the task execution information;

[0007] Execute the sub-process according to the target and select the functional interface corresponding to the text generation task;

[0008] The functional interface is called to execute the text generation task according to the target execution sub-process to obtain the target text corresponding to the text generation task.

[0009] According to a second aspect of the embodiments of this specification, a text generation method is provided, which is applied to an intelligent decision-making component, including:

[0010] Determining a text generation task and task execution information corresponding to the text generation task, wherein the task execution information is used to determine task requirements of the text generation task;

[0011] Determining a target execution sub-process in a preset execution process based on the task execution information;

[0012] Execute the sub-process according to the target and select the functional interface corresponding to the text generation task;

[0013] The functional interface is called to execute the text generation task according to the target execution sub-process to obtain the target news text corresponding to the text generation task.

[0014] According to a third aspect of the embodiments of this specification, a text generation method is provided, which is applied to an intelligent decision-making component, including:

[0015] Receive a text generation task, and determine task execution information corresponding to the text generation task, wherein the task execution information is used to determine task requirements of the text generation task;

[0016] The task execution information is input into a text generation model to obtain a target text corresponding to the text generation task output by the text generation model based on the task execution information, wherein the target text is generated by the text generation model according to the above-mentioned text generation method.

[0017] According to a fourth aspect of the embodiments of this specification, a text generation method is provided, which is applied to a cloud-side device, including:

[0018] receiving a text generation task sent by a terminal-side device, and determining task execution information corresponding to the text generation task, wherein the task execution information is used to determine a task requirement of the text generation task;

[0019] The task execution information is input into a text generation model to obtain a target text corresponding to the text generation task output by the text generation model based on the task execution information, and the target text is sent to the terminal device, wherein the target text is generated by the text generation model according to the above-mentioned text generation method.

[0020] According to a fifth aspect of the embodiments of this specification, a text generation device is provided, which is applied to an intelligent decision-making component, including:

[0021] A first determining module is configured to determine a text generation task and task execution information corresponding to the text generation task, wherein the task execution information is used to determine a task requirement of the text generation task;

[0022] A second determining module is configured to determine a target execution sub-process in a preset execution process according to the task execution information;

[0023] A selection module is configured to execute a sub-process according to the target and select a functional interface corresponding to the text generation task;

[0024] The calling module is configured to call the functional interface to execute the text generation task according to the target execution sub-process, and obtain the target text corresponding to the text generation task.

[0025] According to a sixth aspect of the embodiments of this specification, a text generation device is provided, which is applied to an intelligent decision-making component, including:

[0026] A first determining module is configured to determine a text generation task and task execution information corresponding to the text generation task, wherein the task execution information is used to determine a task requirement of the text generation task;

[0027] A second determining module is configured to determine a target execution sub-process in a preset execution process according to the task execution information;

[0028] A selection module is configured to execute a sub-process according to the target and select a functional interface corresponding to the text generation task;

[0029] The calling module is configured to call the functional interface to execute the text generation task according to the target execution sub-process, and obtain the target news text corresponding to the text generation task.

[0030] According to a seventh aspect of the embodiments of this specification, a text generation device is provided, which is applied to an intelligent decision-making component, including:

[0031] A receiving module is configured to receive a text generation task and determine task execution information corresponding to the text generation task, wherein the task execution information is used to determine a task requirement of the text generation task;

[0032] An input module is configured to input the task execution information into a text generation model to obtain a target text corresponding to the text generation task output by the text generation model based on the task execution information, wherein the target text is generated by the text generation model according to the above-mentioned text generation method.

[0033] According to an eighth aspect of the embodiments of this specification, a text generation apparatus is provided, which is applied to a cloud-side device, including:

[0034] a receiving module configured to receive a text generation task sent by a terminal-side device and determine task execution information corresponding to the text generation task, wherein the task execution information is used to determine a task requirement of the text generation task;

[0035] The sending module is configured to input the task execution information into a text generation model, obtain the target text corresponding to the text generation task output by the text generation model based on the task execution information, and send the target text to the terminal device, wherein the target text is generated by the text generation model according to the above-mentioned text generation method.

[0036] According to a ninth aspect of the embodiments of this specification, a text generation system is provided, the system comprising a client and a server, wherein:

[0037] The client is used to send a text generation task to the server;

[0038] The server is used to determine the task execution information corresponding to the text generation task, wherein the task execution information is used to determine the task requirements of the text generation task, determine the target execution sub-process in the preset execution process according to the task execution information, select the functional interface corresponding to the text generation task according to the target execution sub-process, call the functional interface to execute the text generation task according to the target execution sub-process, obtain the target text corresponding to the text generation task, and send the target text to the client.

[0039] According to a tenth aspect of the embodiments of this specification, there is provided a computing device, including: a memory and a processor;

[0040] The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the above-mentioned text generation method are implemented.

[0041] According to an eleventh aspect of the embodiments of this specification, a computer-readable storage medium is provided, which stores computer-executable instructions, and when the instructions are executed by a processor, the steps of the above-mentioned text generation method are implemented.

[0042] According to a twelfth aspect of the embodiments of this specification, a computer program is provided, comprising a computer program or instructions, which implement the steps of the above-mentioned text generation method when executed by a processor.

[0043] One embodiment of this specification implements the determination of a text generation task and the task execution information corresponding to the text generation task through an intelligent decision-making component, and determines a target execution sub-process in a preset execution process based on the task execution information, so that the intelligent decision-making interface automatically decides to complete the execution process of the text generation task, thereby reducing labor costs and improving text generation efficiency. Subsequently, according to the target execution sub-process, the functional interface corresponding to the text generation task is selected from the preset interface set corresponding to the target field, and the functional interface is called to execute the text generation task according to the target execution sub-process, and the target text corresponding to the text generation task is obtained. By selecting the functional interface corresponding to the text generation task according to the target execution sub-process and calling the functional interface to complete the text generation task, the cost overhead caused by manual processing is further reduced, and the complexity of text generation is reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is an architectural diagram of a text generation system provided by one embodiment of this specification;

[0045] Figure 2 This is a scenario diagram of a text generation method provided by an embodiment of this specification;

[0046] Figure 3 is a flowchart of a text generation method provided by one embodiment of this specification;

[0047] Figure 4 This is a flowchart of a text generation method according to an embodiment of the present invention;

[0048] Figure 5 is a flowchart of another text generation method provided by an embodiment of this specification;

[0049] Figure 6 is a flowchart of another text generation method provided by an embodiment of this specification;

[0050] Figure 7 is a flowchart of another text generation method provided by an embodiment of this specification;

[0051] Figure 8 This is a structural diagram of a text generation device provided by an embodiment of this specification;

[0052] Figure 9 This is a schematic diagram of the structure of another text generation device provided by an embodiment of this specification;

[0053] Figure 10 This is a schematic diagram of the structure of another text generation device provided by an embodiment of this specification;

[0054] Figure 11This is a schematic diagram of the structure of another text generation device provided by an embodiment of this specification;

[0055] Figure 12 This is a structural block diagram of a computing device provided by one embodiment of this specification. DETAILED DESCRIPTION

[0056] The following description sets forth many specific details to facilitate a thorough understanding of this specification. However, this specification can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the scope of this specification. Therefore, this specification is not limited to the specific implementations disclosed below.

[0057] The terms used in one or more embodiments of this specification are for the purpose of describing specific embodiments only and are not intended to limit one or more embodiments of this specification. The singular forms "a," "the," and "the" used in one or more embodiments of this specification and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in one or more embodiments of this specification refers to and includes any or all possible combinations of one or more associated listed items.

[0058] It should be understood that although the terms first, second, etc. may be used to describe various information in one or more embodiments of this specification, such information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of one or more embodiments of this specification, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0059] In addition, it should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in one or more embodiments of this specification are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0060] In one or more embodiments of this specification, a large model refers to a deep learning model with large-scale model parameters, typically containing hundreds of millions, tens of billions, hundreds of billions, trillions, or even more than ten trillion model parameters. A large model can also be called a cornerstone model / foundation model. It is pre-trained on a large-scale unlabeled corpus to produce a pre-trained model with more than 100 million parameters. This model can adapt to a wide range of downstream tasks and has good generalization capabilities, such as a large language model (LLM) and a multi-modal pre-training model.

[0061] When large models are used in practice, only a small number of samples are needed to fine-tune the pre-trained model and it can be applied to different tasks. Large models can be widely used in natural language processing (NLP), computer vision and other fields. Specifically, they can be applied to computer vision tasks such as visual question answering (VQA), image caption (IC), and image generation, as well as natural language processing tasks such as text-based sentiment classification, text summary generation, and machine translation. The main application scenarios of large models include digital assistants, intelligent robots, search, online education, office software, e-commerce, and intelligent design.

[0062] First, the terms involved in one or more embodiments of this specification are explained.

[0063] Agent: An entity or program that can perceive the environment, understand input information, and make decisions and actions based on preset goals and rules. The agent mentioned in one or more embodiments of this specification can be understood as an AI agent (AI agent), that is, a role that uses AI to make automatic decisions when solving a problem. Decisions mainly include specific planning and execution, and the next decision is usually determined based on the feedback and current status of each decision.

[0064] Knowledge illusion: The model is overconfident about the knowledge it has learned previously. As a result, when predicting results, the model makes predictions based on the previously learned knowledge. For example, a question-answering model makes question-answer predictions based on the previously learned knowledge, which leads to factual errors.

[0065] In the current media news creation landscape, the news article creation process is highly specialized and rigorous, placing extremely high demands on the authenticity, objectivity, professionalism, logic, depth, and innovation of the manuscript. Furthermore, this process often requires a professional editor to manually conduct multiple steps, including analysis, data collection and organization, contextualization, writing, proofreading, and review, to complete a single manuscript. This is inefficient and labor-intensive. While current methods that directly create news articles based on large models can produce news, they cannot guarantee objectivity, authenticity, professionalism, and depth, and are often prone to problems such as knowledge illusions and unclear logic. For example, large-model-based news writing methods directly leverage the general generative capabilities of large models, relying on prior knowledge learned during pre-training, to write on a user-given topic without relying on any other reference materials. While this method can generate news articles, it cannot guarantee objectivity, authenticity, professionalism, and depth. Another example is news writing based on retrieval enhancement. This method introduces retrieval information sources, converts user questions into query keywords, and then directly gives the retrieval results to the model. The model writes news based on the retrieval results. This method can improve the factuality and reliability of the news to a certain extent and avoid the problem of excessive hallucinations in the model. However, this method only enhances information based on the retrieved content without further screening the information. In extreme cases, it may introduce more noise. In addition, this method still lacks the rationality of search material selection, content depth and authenticity, which may cause the generated results to become worse.

[0066] Based on this, a text generation method is provided in this specification. This specification also involves a text generation device, a computing device, a computer-readable storage medium, and a computer program product, which are described in detail one by one in the following embodiments.

[0067] See also Figure 1 , Figure 1 The following is an architecture diagram of a text generation system provided by an embodiment of the present specification. The text generation system may include a client 100 and a server 200;

[0068] The client 100 is used to send a text generation task to the server 200;

[0069] The server 200 is configured to execute a text generation task sent by a client, including determining a text generation task and task execution information corresponding to the text generation task, wherein the task execution information is used to determine the task requirements of the text generation task; determining a target execution sub-process in a preset execution process based on the task execution information; selecting a functional interface corresponding to the text generation task based on the target execution sub-process; calling the functional interface to execute the text generation task according to the target execution sub-process to obtain a target text corresponding to the text generation task; and sending the target text to the client 100.

[0070] The client 100 is further configured to receive the target text sent by the server 200 .

[0071] By applying the solution of the embodiments of this specification, an intelligent decision-making component, namely an AI Agent, can be deployed in the server 200 to execute the text generation method through the intelligent decision-making component. In a specific application scenario, the client 100 has a text generation demand, such as the demand for generating and creating a press release for a hot topic of a certain phenomenon, or the demand for creating a commentary article for a certain activity. In this case, the client 100 will generate a text generation task related to the field of the manuscript and send the task to the server 200. The server 200 can provide text generation services to the client 100, such as providing text generation functions to users through certain applications or software. When the server 200 receives a text generation task, it will determine the task execution information corresponding to the text generation task, and then determine the user's writing needs, writing topics and other information based on the task execution information. According to the task execution information, the target execution sub-process is determined in the preset execution process. According to the target execution sub-process, the functional interface that needs to be called subsequently is selected from the preset interface set of the associated target field. By calling the functional interface, the text generation task is executed according to the target execution sub-process, thereby generating the target text expected by the user. After generating the target text, the server 200 can return the target text to the client 100.

[0072] The text generation system may include multiple clients 100 and multiple servers 200. The clients 100 may be referred to as client-side devices, and the servers 200 may be referred to as cloud-side devices. The multiple clients 100 may establish a communication connection through the servers 200. In a text generation scenario, the servers 200 provide text generation services between the multiple clients 100. The multiple clients 100 may act as either senders or receivers, communicating through the servers 200.

[0073] Users can interact with the server 200 through the client 100 to receive data sent by other clients 100, or send data to other clients 100, etc. In the text generation scenario, the user can publish a data stream to the server 200 through the client 100, and the server 200 can generate a target text based on the data stream and push the target text to other clients with which communication has been established.

[0074] The client 100 and the server 200 are connected via a network. The network provides a medium for the communication link between the client 100 and the server 200. The network can include various connection types, such as wired or wireless communication links or fiber optic cables. The data transmitted by the client 100 may need to be encoded, transcoded, compressed, or other processing before being released to the server 200.

[0075] The client 100 can be a browser, an application (Application, APP), or a web application such as an H5 (HyperText Markup Language 5, Hypertext Markup Language 5) application, or a light application (also known as a mini-program, a lightweight application) or a cloud application. The client 100 can be based on the software development kit (SDK) of the corresponding service provided by the server 200, such as developed based on the real-time communication (RTC) SDK. The client 100 can be deployed in an electronic device and needs to rely on the device to run or certain APPs in the device to run. For example, the electronic device can have a display screen and support information browsing, such as a personal mobile terminal such as a mobile phone, a tablet computer, a personal computer, etc. Various other types of applications can also be configured in the electronic device, such as human-computer dialogue applications, model training applications, text processing applications, web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0076] The server 200 may include servers that provide various services, such as servers that provide communication services to multiple clients, servers that support background training for models used on clients, and servers that process data sent by clients. It should be noted that the server 200 can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. The server can also be a server in a distributed system, or a server that integrates a blockchain. The server can also be 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, content delivery networks (CDNs), and big data and artificial intelligence platforms, or an intelligent cloud computing server or intelligent cloud host with artificial intelligence technology.

[0077] It is worth noting that the text generation method provided in the embodiments of this specification is generally executed by the server. However, in other embodiments of this specification, the client may also have similar functions to the server to execute the text generation method provided in the embodiments of this specification. In other embodiments, the text generation method provided in the embodiments of this specification may also be executed jointly by the client and the server.

[0078] See also Figure 2 , Figure 2 A scenario diagram of a text generation method provided according to an embodiment of the present specification is shown, wherein, in a text generation scenario, an intelligent decision-making component receives a text generation task for a target field. The intelligent decision-making component is an AI agent. The intelligent decision-making component determines the task execution information corresponding to the text generation task, and determines the target execution sub-process corresponding to the text generation task in the preset execution process based on the task execution information. Figure 2 The preset execution process is "Analysis - Research - Material Organization - Article Writing - Manuscript Review - Other Assistance". These process nodes are combined to generate the execution sub-process. Figure 2 The process nodes "Analysis - Research - Material Organization - Article Writing" contained in the dotted box are combined to generate. The intelligent decision component automatically determines the execution process node corresponding to the text generation task based on the text generation task. Subsequently, the functional interface for executing the text generation task can be determined in the preset interface set based on the determined execution sub-process. Figure 2The preset interface set includes "analysis interface, search interface, organization interface, generation interface, proofreading interface, and auxiliary interface". Each interface can correspond to a process node, and the corresponding functional component can be called through the functional interface. Specific functional components may include "analysis module, research module, material organization module, article writing module, manuscript proofreading module, and other tool modules". The analysis module can be called through the analysis interface, the research module can be called through the search interface, the material organization module can be called through the organization interface, the article writing module can be called through the generation interface, the manuscript proofreading module can be called through the proofreading interface, and other tool modules can be called through the auxiliary interface. Therefore, after determining the execution process node, the functional interface for executing the text generation task can be selected from the preset interface set based on the execution process node, such as Figure 2 The functional interfaces in the dotted box are "analysis interface, search interface, organization interface, and generation interface". The subsequent intelligent decision-making components will call these functional interfaces according to the execution sub-process. By calling these functional interfaces, the functional components corresponding to the functional interfaces are used to perform the corresponding functions, thereby completing the text generation task and obtaining the target text corresponding to the text generation task. The intelligent decision-making body automatically selects the execution process of text generation, thereby reducing labor costs and dividing the text generation process into multiple steps. Each step has a corresponding functional interface, so that the subsequent task can be completed by calling the functional interface, which improves the efficiency of task execution and generates higher quality text.

[0079] See also Figure 3 , Figure 3 A flowchart of a text generation method provided according to an embodiment of this specification is shown. The method is applied to an intelligent decision-making component and specifically includes the following steps.

[0080] Step 302: Determine a text generation task and task execution information corresponding to the text generation task, wherein the task execution information is used to determine the task requirements of the text generation task.

[0081] In actual applications, when creating texts such as news releases and articles, users can realize automatic creation of texts by using the text generation method applied to the intelligent decision-making component provided in this specification, thereby reducing the manpower and time costs. The intelligent decision-making component can be deployed on a terminal that provides a text generation service. The user uses the text generation service through an application on the client, thereby calling the intelligent decision-making component to perform text generation. In the process of the user calling the intelligent decision-making component, a text generation task will be generated and sent to the intelligent decision-making component. After the intelligent decision-making component determines the text generation task for the target field, it will determine the task execution information corresponding to the text generation task. The task execution information may include instructions entered by the user, such as "Please help me generate a news release about "hot search events"". In actual applications, the user's expected goals can be determined based on the task execution information, that is, the user's needs can be determined, such as whether the user wants to generate a news release or to refine the news release. Different processes can be determined according to different needs, and the relevant API interfaces can be called to use the corresponding functional services. The respective processes are completed through the functional services to improve processing efficiency.

[0082] In different fields, due to differences in language style and wording, the text generated by text generation tasks in different fields also has different styles. To improve the professionalism of the generated text and make it consistent with the field to which it belongs, the user can subsequently select the interface corresponding to the target field and use the services provided by the functional components corresponding to the interface to perform the task. The target field can be understood as the field to which the text the user wants to generate belongs. For example, if the user wants to generate a news article, the target field is the news field; if the user wants to generate an entertainment magazine article, the target field is the magazine article field; if the user wants to generate a topic article for an online platform, the target field is the online field. It should be noted that the functional components of the functional interfaces corresponding to different fields can be trained based on training data from different fields. A text generation task is a task instruction for generating text. A text generation task can be generated by the user using the text generation service and sent to the intelligent decision-making component. The intelligent decision-making component can be understood as the AI ​​agent in the text generation service that replaces professional editors in the step-by-step thinking and decision-making process in the text generation process.

[0083] In a specific embodiment of this specification, when a user wants to use a text generation service to automatically generate a news report text about a hot search event, the text generation method provided by this application can be used for automatic generation. A specific user can generate a text generation task by deploying an application with a text generation function, and send it to the server that provides the text generation service. The intelligent decision-making component on the server determines the text generation task and determines the task execution information corresponding to the task. The task execution information can be understood as the instruction information entered by the user when generating the text generation task, including user needs, text topics and other information. The intelligent decision-making component determines the task requirements of the text generation task through the task execution information, that is, it understands the user's relevant requirements, intentions, etc. for the generated text this time, so as to facilitate the subsequent generation of article texts that are more in line with user expectations.

[0084] Step 304: Determine a target execution sub-process in a preset execution process according to the task execution information.

[0085] In actual application, after determining the task execution information corresponding to the text generation task, the target execution sub-process can be determined in the preset execution process according to the task execution information. The preset execution process can be understood as a process for processing text generation tasks that is set in advance. The preset execution process is a general execution process. In one embodiment of this specification, the preset execution process includes analysis, research, material organization, article writing, manuscript verification, and other auxiliary nodes. By dividing text generation into multiple execution processes, the article creation process is broken down into multiple atomic steps, which can better generate more professional articles. The target execution sub-process is the execution process selected from the preset execution process based on the task execution information. Different user needs can be determined for different task execution information. For example, if some users only want to search for some material data about the current hot spot, then their ultimate demand goal is to search for materials. At this time, the preset execution sub-process determined is to organize materials; or if some users want to automatically generate a press release about the current hot spot, then their ultimate demand goal is to generate a press release. At this time, the preset execution sub-process determined is to generate article. Therefore, different target execution sub-processes can be selected from the preset execution process for different task execution information.

[0086] In actual applications, the execution order of the target execution sub-process may not be in the order of the preset execution process. For example, if the user only needs to conduct research and article writing, then only these two process nodes can be selected to form a preset execution process for execution, and the intermediate material organization execution step can be omitted.

[0087] In a specific embodiment of the present specification, citing the above example, the target execution sub-process is determined in the preset execution process according to the task execution information. The task execution information is "Please help me generate a press release about "hot search events"". The preset execution process includes "analysis, research, material organization, article writing, manuscript proofreading, and other assistance". According to the task execution information, the target execution sub-process is determined to be "analysis, research, material organization, and article writing".

[0088] Furthermore, in order to accurately determine the target execution sub-process, it is necessary to determine the user's required goals based on the task execution information. Specifically, based on the task execution information, the target execution sub-process is determined in the preset execution process, including: determining the task goal corresponding to the task execution information, selecting the process node corresponding to the task goal in the preset execution process; and generating the target execution sub-process based on the process node.

[0089] Among them, the task goal corresponding to the task execution information can be understood as the user's desired demand goal. For example, if the user expects to automatically generate a press release, the task goal is to generate a press release; if the user expects to automatically generate comments about the press release, the task goal is to generate comment text. Therefore, according to the determined task goal, a process node that meets the user's expected needs can be selected in the preset execution process, and then a target execution sub-process can be generated based on the process node. The process node is the node corresponding to each execution process contained in the preset execution process. After selecting the process node corresponding to the task goal, the selected process node can be generated into a target execution process. In special cases, if the number of process nodes is one, the process node can be directly used as the target execution sub-process.

[0090] In actual applications, when selecting process nodes in the preset execution process based on the task goal corresponding to the task execution information, the selection can be made according to certain rules based on the task goal. For example, the preset execution process includes process nodes such as research, material organization, and article writing. When the task goal is article writing, you may choose only article writing or you may choose "research-material organization-article writing". The specific selection method needs to be determined based on the specific content of the task execution information. If the task goal determined according to the task execution information is that the user only wants to proofread the article, then other auxiliary process nodes can be selected. For example, if the task goal determined is that the user wants to generate an article, then the analysis, research, material organization, and article writing process nodes can be selected. Therefore, in different situations, you can select a single process node, or you can select multiple process nodes in the execution order. The specific selection method is determined according to the specific situation.

[0091] In a specific embodiment of the present specification, referring to the above example, the task goal corresponding to the task execution information is determined. The task goal is to generate a press release. Then, according to the execution process specified in the preset execution process, the selected process nodes include analysis, research, material organization, and article generation. Then, these process nodes are combined in the execution order to generate the target execution sub-process "analysis-research-material organization-article generation".

[0092] Based on this, by analyzing the user's expected demand goals based on the task execution information, the process nodes that need to be executed can be accurately screened out according to the task goals, thereby generating a target execution sub-process corresponding to the task execution information, which is convenient for subsequent calling the corresponding interface based on the target execution sub-process to execute the text generation task.

[0093] Step 306: Execute a sub-process according to the target and select a functional interface corresponding to the text generation task.

[0094] In actual applications, each process node in the preset execution process will correspond to a preset interface in the preset interface set of the target domain. The preset interface set can be understood as the set of interfaces corresponding to each process node in the preset execution process. Each interface in the preset interface set corresponds to a functional component, that is, each functional component provides an external API interface for calling. Through this interface, the functional services provided by the functional component can be used, so that the corresponding functional component can execute the corresponding process node by calling the interface, thereby executing the steps corresponding to the corresponding process node according to calling the preset interface and completing the text generation task. Each preset interface can run a functional component, which is used to execute the corresponding process node to complete the corresponding function.

[0095] In the embodiments of this specification, the preset interface set includes an analysis interface, a search interface, an organization interface, a generation interface, a proofreading interface, and an auxiliary interface. The analysis interface corresponds to the analysis process node, the search interface corresponds to the research process node, the organization interface corresponds to the material organization node, the generation interface corresponds to the article writing node, the proofreading interface corresponds to the manuscript process node, and the auxiliary interface corresponds to other auxiliary nodes. In different target fields, although the preset execution process is the same, the corresponding preset interface sets are different. Since the functional components corresponding to the preset interfaces can use large models to implement functions, the functional components corresponding to the preset interfaces in different fields need to be trained using training data from different fields, so that when performing text generation tasks for the target field later, it is necessary to select a preset interface from the preset interface set corresponding to the target field, and use the selected preset interface as the functional interface used when performing the text generation task, and run its corresponding functional component through the functional interface to complete the task.

[0096] In a specific embodiment of this specification, citing the above example, the target execution sub-process is determined to be "analysis-research-material organization-article writing". According to the target execution sub-process, the functional interfaces corresponding to the text generation task are selected from the preset interface set, including analysis interface, search interface, organization interface, and generation interface.

[0097] Based on this, by selecting the functional interface for executing the text generation task from the preset interface set according to the target execution sub-process, and through the functional interfaces corresponding to multiple execution processes, not only can the text be generated in a more fine-grained manner, reducing the difficulty of text generation, but also the labor input cost can be reduced.

[0098] Furthermore, in order to avoid the subsequent inability to execute the text generation task, it is necessary to select a functional interface corresponding to the text generation task from the preset interface set. Specifically, according to the target execution sub-process, the functional interface corresponding to the text generation task is selected from the preset interface set corresponding to the target field, including: determining the target process node contained in the target execution sub-process, and the node identifier corresponding to the target process node; according to the node identifier, selecting the target functional interface corresponding to the target process node from the preset interface set corresponding to the target field; and using the target functional interface as the functional interface corresponding to the text generation task.

[0099] Among them, the target process node contained in the target execution sub-process can be understood as the process node selected from the target execution sub-process, that is, one is selected from the multiple process nodes corresponding to the target execution sub-process as the target process node, and the node identifier corresponding to the target process node can be understood as the unique identifier corresponding to the process node, which can be used to determine the functional interface corresponding to the process node. Therefore, according to the node identifier, the target functional interface corresponding to the target process node can be selected from the preset interface set.

[0100] In actual applications, the target execution sub-process may include multiple process nodes. The functional interface corresponding to each process node is determined in sequence. In this case, the corresponding functional interface can be selected from a preset interface set based on the node identifier corresponding to each process node, thereby using the target functional interface corresponding to each process node as the functional interface for the text generation task. In specific implementations, the process nodes and functional interfaces can be pre-associated, facilitating the subsequent determination of the target functional interface associated with the target process node based on the node identifier.

[0101] In a specific embodiment of this specification, citing the above example, the target execution sub-process is "Analysis - Research - Material Organization - Article Writing." The target process node is selected as the "Analysis" process node, the node identifier of the target process node is determined, and the target functional interface corresponding to the target process node is selected as the "Analysis Interface" based on the node identifier. Following the above method, the target functional interface corresponding to each target process node in the target execution sub-process is sequentially determined, including the "Analysis Interface, Search Interface, Organization Interface, and Generation Interface." These target functional interfaces are then used as the functional interfaces corresponding to the text generation task.

[0102] Based on this, the target functional interface corresponding to each target process node in the target execution sub-process can be selected according to the node identifier, and the determined target functional interface can be used as the functional interface corresponding to the text generation task, so as to facilitate the subsequent execution of the text generation task through the functional interface.

[0103] Furthermore, the functional interface includes at least one of an analysis interface, a search interface, an organization interface, a generation interface, a proofreading interface and an auxiliary interface.

[0104] In practical applications, the text generation method provided in this specification breaks down the news release creation process into multiple atomic steps, including analysis, research, material organization, article generation, manuscript verification, and other auxiliary steps, to produce more objective, detailed, professional, high-quality, logical, and in-depth news releases. This avoids the lack of authenticity and depth of content in the generated news due to the lack of some core pre-writing steps (such as multi-step search and event excerpting). The execution process of the text generation method provided in this specification can well simulate the process of professional editors writing news manuscripts, improving the objectivity, professionalism, and logic of news releases.

[0105] During specific implementation, in order to meet the step execution requirements of each process, it is necessary to provide corresponding functional interfaces for step execution. The functional component corresponding to the functional interface can be a large model obtained through training with a large amount of training data. The intelligent decision-making component can automatically decide how to combine the steps, thereby calling the corresponding interface. In a specific embodiment of this specification, the functional interface includes an analysis interface, a search interface, an organization interface, a generation interface, a proofreading interface and an auxiliary interface. The analysis interface is mainly used to clarify the subject and requirements of the text generation task, the search interface is mainly used to search for materials related to the subject, the organization interface is mainly used to summarize the searched materials, the generation interface is mainly used to generate articles based on the summarized materials, the proofreading interface is mainly used to proofread the generated articles, and other auxiliary interfaces are mainly used to provide other auxiliary functions for the generated text, such as generating titles, comments, summaries and other functions.

[0106] Based on this, the execution process is realized by running the corresponding functional components through various functional interfaces, which reduces the labor input cost, and through the mutual cooperation of various functional interfaces, text that better meets user needs can be generated.

[0107] Step 308: Call the functional interface to execute the text generation task according to the target execution sub-process to obtain the target text corresponding to the text generation task.

[0108] In actual applications, after determining the functional interface, the intelligent decision-making component can run the functional component by calling the functional interface, and execute the text generation task according to the target execution sub-process, thereby obtaining the target text corresponding to the text generation task. In specific implementation, when the intelligent decision-making component, i.e., the AI ​​agent, calls the functional interface to make the functional component corresponding to the interface perform the task, the AI ​​agent can automatically make a decision based on the execution result. For example, when calling the search interface to search for relevant materials, if the AI ​​agent determines that the currently searched materials are not enough, it can continue to choose to call the search interface to search. After the AI ​​agent determines that it does not need to continue searching, it can continue to call the organization interface to organize the currently searched materials.

[0109] Based on this, the intelligent decision-making component calls the functional interface to execute the text generation task according to the target execution sub-process, so that the target text corresponding to the text generation task can be obtained, thereby achieving the effect of automatically generating the target text and reducing the cost of manual creation input.

[0110] Furthermore, in the case where the AI ​​agent calls the analysis interface, the functional interface is the analysis interface; wherein, calling the functional interface to execute the text generation task according to the target execution sub-process includes: calling the analysis interface to analyze the task execution information to obtain the text generation information corresponding to the text generation task; and outputting the text generation information according to the target execution sub-process.

[0111] In actual applications, when the AI ​​agent calls the analysis interface, it can use the analysis interface to execute the steps corresponding to the analysis process node. The analysis interface is mainly used to analyze the task execution information of the text generation task, and analyze the specific writing theme and writing requirements, such as the required writing style, length, position, genre, perspective, etc. The analysis interface mainly implements two functions, namely creative theme extraction and writing requirement analysis. Creative theme extraction is to extract the specific writing theme in the information, such as news theme, article theme, etc.; writing requirement analysis is to analyze the specific writing needs and requirements in the information, such as the content style and content length specified by the user.

[0112] Among them, text generation information can be understood as the information required when generating text. Text generation information includes subject information and writing information. Subject information is the creation theme specified by the user, such as a news hotspot, an event theme or a thing theme, etc. Writing information is the creation requirements or needs specified by the user when generating text. For example, if the user points out his or her views on an event, it is necessary to create from the user's perspective, or if the user requires the length of the creation content, then the subsequent text generation needs to be generated according to a certain length. After analyzing the task execution information through the analysis interface and obtaining the text generation information, the text generation information can be output according to the target execution sub-process. In specific implementation, the analysis interface will send the text generation information to the search interface so that the search interface can use the text generation information to search for related materials.

[0113] In a specific embodiment of the present specification, citing the above example, the AI ​​agent, i.e., the intelligent decision-making component, calls the analysis interface to analyze the task execution information, determines that the subject information corresponding to the text generation task is a "hot search event", and determines that the text generation information is a "press release", and then outputs the determined text generation information to the next called functional interface in the target execution process.

[0114] Based on this, by calling the analysis interface, the specific creative theme and creative requirements can be accurately analyzed from the information input by the user, so that a text that meets the user's requirements can be generated subsequently.

[0115] Furthermore, when the AI ​​agent calls the search interface, the functional interface is the search interface; wherein, calling the functional interface to execute the text generation task according to the target execution sub-process includes: calling the search interface to perform search processing based on the text generation information to obtain material data associated with the text generation information; and outputting the material data according to the target execution sub-process.

[0116] In actual applications, after obtaining the subject information and text generation information through the analysis interface, the analysis interface can input the text generation information into the search interface. When the AI ​​agent calls the search interface, it can search for material data related to the text generation information and other materials related to the current search content through the search interface.

[0117] The material data associated with the subject information can be understood as the material data that the search interface searches for based on the subject information. The material data is related to the subject information. For example, if the subject information is a song, the material data searched for can be news reports related to the song. The material data can also be related to writing information. For example, if the search is based on the user's views on an event, the material data searched for can be articles related to the event that are the same or similar to the user's views. After obtaining the material data associated with the subject information, the search interface can output the material data to the next functional interface according to the target execution sub-process.

[0118] In practice, the search interface is divided into private and public searches. Private searches can be used to search internal databases and resources, such as those built independently by some companies. It should be noted that private searches are conducted with the authorization of the resource owner, ensuring the privacy and security of internal data. Public searches search for relevant resources on the public network.

[0119] In a specific embodiment of the present specification, citing the above example, the AI ​​agent calls the search interface to search for materials about "hot search events", and outputs the materials related to the "hot search events" obtained from the search to the next called functional interface in the target execution process.

[0120] Based on this, the search interface can be used to search for relevant materials related to the information generated by the associated text, ensuring that when the text is subsequently generated, the factuality and reliability of the text are improved and the problem of excessive hallucination is avoided.

[0121] Furthermore, in order to ensure that the search materials meet the requirements of text generation, the intelligent decision-making component can determine whether to continue searching. Specifically, after obtaining the material data associated with the text generation information, it also includes: determining the execution strategy of the search interface based on the material data; when the execution strategy is a continue execution strategy, calling the search interface to generate search keywords based on the material data, using the search keywords as the subject information, and calling the search interface to continue the step of searching based on the subject information; when the execution strategy is a stop execution strategy, calling the search interface to continue the step of outputting the material data according to the target execution sub-process.

[0122] The execution strategy of the search interface can be understood as the strategy used by the intelligent decision-making component to determine whether to continue calling the search interface based on the current search content. If the current search material meets the preset conditions, the intelligent decision-making component will automatically select the execution strategy as the stop execution strategy; if the current search material does not meet the preset conditions, the intelligent decision-making component will automatically select the execution strategy as the continue execution strategy. The stop execution strategy means that the next execution process can be continued for the current search material, and there is no need to continue calling the search interface for search; the continue execution strategy means that the next execution process cannot be executed for the current search material, and it is necessary to continue calling the search interface for search.

[0123] In actual applications, after the AI ​​agent obtains the search results, i.e., material data, fed back by the calling interface, it can make a judgment on the material data to determine whether it is necessary to continue searching, and then select an execution strategy based on preset conditions. The preset conditions can be set in advance according to actual conditions, such as whether the material data obtained by the search meets the preset data volume, or whether the material data obtained by the search includes the expected content. In specific implementation, the search interface also includes two functions: association analysis and multi-step search. Association analysis is to analyze the search results and core information, and organize and expand new search keywords. The search keywords are the search terms used in the next round of search. Multi-step search is to further search for related materials based on the expanded search keywords. The AI ​​agent decides whether to perform multiple searches and how many searches to perform. In the multi-step search process, if the AI ​​agent has determined that the related materials obtained by the current search have met the conditions, then the related materials obtained by each search can be used as material data of the related subject information and output.

[0124] In a specific embodiment of this specification, citing the above example, after the search interface performs a material search for "hot search events," material data associated with the "hot search events," such as news reports on the "hot search events," is obtained. If the AI ​​agent determines that the execution strategy is to continue execution, the search keyword can be expanded based on the news reports obtained from the search, and the search interface can be called to search based on the expanded search keyword until the conditions for stopping the search are met. If the execution strategy is to stop execution, all material data obtained from the previous searches are output as material data associated with the "hot search events."

[0125] Based on this, searching for materials based on subject information through the search interface can not only obtain material data related to the subject information, facilitating subsequent text generation based on the material data and improving the authenticity and reliability of the text, but also automatically judge multi-step searches based on AI intelligent agents to ensure that the searched materials meet the generation requirements.

[0126] Furthermore, when the AI ​​agent calls the organization interface, the functional interface is the organization interface; wherein, calling the functional interface to execute the text generation task according to the target execution sub-process includes: calling the organization interface to summarize and process the material data according to preset organization rules to obtain target material data in the material data, wherein the preset organization rules are used to determine the organization conditions for the material data; generating text association information based on the target material data, and outputting the text association information according to the target execution sub-process.

[0127] The preset organization rules can be understood as rules for organizing material data that are set in advance for the organization interface. Based on the preset organization rules, the organization conditions for the material data can be determined. The organization conditions can include summarizing, organizing, and arranging the searched material data, including article recall sorting, outline generation, event extraction, event excerpting, etc. After calling the organization interface to summarize and process the material data, the target material data in the material data can be recalled. In other words, the organization interface can be used to screen the searched material data to improve the data quality of the material data.

[0128] In practical applications, article recall sorting includes recalling article materials related to the topic and re-sorting the article materials from the perspective of relevance. Since the material data obtained by the search interface is related to the topic information for the search interface, in fact, during the search process, search deviation may occur due to the expansion of keywords. In this case, the organization interface can be used to sort the search materials by article recall, that is, to recall materials related to the topic information from the material data, eliminate some redundant noise information, improve the quality of the material data, and sort the material data from the perspective of relevance, so that when the text is generated later, the key material data can be determined according to the sorting order, thereby improving the quality of the text generation.

[0129] In specific implementation, after determining the target material data, the organization interface can generate text-related information based on the target material data. Text-related information is information related to the search material, and text-related information includes outlines, summaries, and descriptions. Therefore, the organization interface also includes three functions: outline generation, event extraction, and event excerpting. Outline generation is to generate an outline of the text based on existing materials before the text is generated, such as organizing the context of the news before writing a press release; event extraction is to extract multiple events related to the theme and outline from the search material and summarize them, such as using a sentence to summarize multiple events related to the theme and outline; event excerpting is to give a brief, truthful, objective, and detailed retelling of each extracted event.

[0130] In a specific embodiment of this specification, the AI ​​agent calls an organization interface to aggregate and process the searched material data, i.e., to recall and sort the articles, obtaining the target material data. It then generates an outline, extracts events, and summarizes them, obtaining text-related information. This information is then input into the functional interface corresponding to the next process node in the target execution sub-process.

[0131] Based on this, the search materials can be further screened through the organizational interface to improve the relevance and authenticity of the search materials, ensure that the subsequently generated text is authentic and reliable, and generate text-related information for the screened material data, so that when subsequently creating text, it can be based on the text-related information, ensuring that the generated content is more objective and professional.

[0132] Furthermore, when the AI ​​agent calls the generation interface, the functional interface is the generation interface; wherein, calling the functional interface to execute the text generation task according to the target execution sub-process includes: calling the generation interface to perform text generation processing according to the text association information to obtain the initial text corresponding to the text generation task; outputting the initial text according to the target execution sub-process.

[0133] The text generation process based on text association information can be understood as the generation interface creating text based on the text association information generated by the organization interface. The generation interface primarily completes text creation based on text association information. The initial text can be understood as the text generated by the text interface. The initial text can then be proofread to obtain the target text, or the initial text can be used directly as the target text without proofreading.

[0134] In actual applications, the generation interface completes the creation of text based on organized and collated materials such as outlines and events. The generation interface can also perform text translation based on the generated text to generate initial texts in different languages, so that it can be subsequently applied to different text publishing platforms. In specific implementation, when the initial text is obtained after text generation, the initial text can be directly output as the target text of the text generation task. At this time, the generation interface can output the initial text as the target text and feed it back to the AI ​​agent, and the AI ​​agent will feed back the target text of the text generation task to the user. The generation interface can also output the initial text to the functional interface corresponding to the next process node to facilitate further processing of the initial text.

[0135] In a specific embodiment of the present specification, citing the above example, the generation interface performs text generation processing based on text association information, generates a press release about the "hot search event", i.e., the initial text, and outputs the initial text to the next functional interface, i.e., the proofreading interface, for manuscript proofreading, so as to obtain a target text that better meets user needs.

[0136] Based on this, the text creation of the text generation task can be completed through the generation interface, thereby realizing automatic text generation and reducing labor costs.

[0137] Furthermore, when the AI ​​agent calls the proofreading interface, the functional interface is the proofreading interface; wherein, calling the functional interface to execute the text generation task according to the target execution sub-process includes: calling the proofreading interface to adjust the initial text based on the text generation information according to preset proofreading rules to obtain the target text corresponding to the text generation task, wherein the preset proofreading rules are used to determine the proofreading conditions for the text generation information; and outputting the target text according to the target execution sub-process.

[0138] In actual applications, when the text generation task is to generate a press release and proofread it, the target execution sub-process determined at this time includes the manuscript proofreading process, and the proofreading interface needs to be called at this time to implement the proofreading of the manuscript. Among them, the preset proofreading rules can be rules set in advance for proofreading texts. According to the preset proofreading rules, the proofreading conditions for the text generation information can be determined. The proofreading conditions are the requirements for proofreading the text, such as detecting formal errors in the text, detecting whether the length meets the requirements, etc. The initial text is adjusted according to the preset proofreading rules, and the proofread target text can be obtained. During the adjustment process, in order to make the target text meet the user's needs, the initial text can be adjusted based on the text generation information input by the user.

[0139] In practice, the proofreading interface performs secondary editing, review, and proofreading on the generated initial text, ensuring that the content and style of the initial text are more reasonable and the text is more appropriate, and preventing content errors. The proofreading interface's functions include rewriting, continuation, expansion, abbreviation, content review, and content proofreading. In the scenario of generating articles, rewriting means rewriting the expression of the article, such as rewriting it into other language styles such as serious, relaxed, etc., or adapting it to the corresponding features of different publishing platforms such as entertainment magazine platforms, popular science magazine platforms, etc.; continuation means that the AI ​​intelligent body automatically determines whether the article is complete. If it is not complete, the proofreading interface is called to supplement the content at the end of the article; expansion means that the AI ​​intelligent body automatically determines whether the article is detailed. If it is not detailed, the proofreading interface will be called to make appropriate supplements in the article; abbreviation means that the AI ​​intelligent body automatically determines whether the article is redundant. If it is redundant, the proofreading interface will be called to delete the content in the article; content review means that the AI ​​intelligent body determines whether the article contains unhealthy or illegal content. At this time, the proofreading interface can be called to rewrite the article or directly return to the generation interface for rewriting; content proofreading means modifying formal errors in the article, such as typos, entity errors, factual errors, etc.

[0140] Based on this, the proofreading interface can be used to further optimize and adjust the initial text generated by the generation interface to ensure that the generated target text is more objective, detailed, professional, high-quality, logical, and in-depth.

[0141] Furthermore, in the case where the AI ​​agent calls an auxiliary interface, the functional interface is the auxiliary interface; wherein, calling the functional interface to execute the text generation task according to the target execution sub-process includes: calling the auxiliary interface to extract information from the target text to obtain auxiliary text associated with the target text; and outputting the auxiliary text according to the target execution sub-process.

[0142] Among them, the auxiliary interface is used when the AI ​​agent or human judgment needs to perform some auxiliary capabilities. For example, the auxiliary interface can extract some keywords and abstracts for the article, or generate a title, and generate comments for the article. By extracting information from the target text through the auxiliary interface, auxiliary text related to the target text can be obtained. The auxiliary text can include text keywords, abstracts, titles, comments, etc. Subsequent calls to the auxiliary interface can feedback the auxiliary text to the user.

[0143] In actual applications, the auxiliary interface can realize functions including keyword extraction, summary generation, title generation and comment generation. Keyword extraction is to extract keywords that can represent the core idea of ​​the article from the article; summary generation is to generate a summary overview that can represent the core idea of ​​the article; title generation is to generate a news headline that can represent the core idea of ​​the article; comment generation is to generate relevant comments around the article content, such as netizen comments, professional reviews, etc.

[0144] Based on this, some auxiliary functions of related articles can be completed through the auxiliary interface, thereby providing users with relevant information related to the target text, improving the richness of the generated text, and reducing the cost of subsequent manual analysis and information extraction of the target text.

[0145] This specification provides a text generation method, which is applied to an intelligent decision-making component, including determining a text generation task and task execution information corresponding to the text generation task, wherein the task execution information is used to determine the task requirements of the text generation task; determining a target execution sub-process in a preset execution process according to the task execution information; selecting a functional interface corresponding to the text generation task from a preset interface set corresponding to the target field according to the target execution sub-process; calling the functional interface to execute the text generation task according to the target execution sub-process, and obtaining the target text corresponding to the text generation task. The method realizes determining the text generation task and the task execution information corresponding to the text generation task by the intelligent decision-making component, determining the target execution sub-process in the preset execution process according to the task execution information, so that the intelligent decision-making component automatically decides to complete the execution process of the text generation task, reduces labor costs, and improves text generation efficiency. Subsequently, according to the target execution sub-process, the functional interface corresponding to the text generation task is selected from the preset interface set corresponding to the target field, and the functional interface is called to execute the text generation task according to the target execution sub-process, and the target text corresponding to the text generation task is obtained. By selecting the functional interface corresponding to the text generation task according to the target execution sub-process and calling the functional interface to complete the text generation task, the cost overhead caused by manual processing is further reduced and the complexity of text generation is reduced.

[0146] The following combined Figure 4 , taking the application of the text generation method provided in this specification in the news field as an example, the text generation method is further explained. Figure 4 A flowchart of a text generation method provided in an embodiment of the present specification is shown, which specifically includes the following steps.

[0147] Step 402: Determine a text generation task and task execution information corresponding to the text generation task.

[0148] In a specific embodiment of this specification, the target field is the news field, and the text generation task is a task issued by the user to generate a news release for hot event A. The user generates the text generation task through a terminal application. After the AI ​​agent receives the text generation task, it determines the corresponding task execution information, which may include the user's needs.

[0149] Step 404: Determine a target execution sub-process in the execution process according to the task execution information.

[0150] In a specific embodiment of the present specification, it is determined that the task goal corresponding to the task execution information is to create a press release, and the process node corresponding to the task goal is selected in the preset execution process "Analysis-Research-Material Organization-Article Writing-Manuscript Verification-Other Auxiliary". The selected process nodes include "Analysis, Research, Material Organization, Article Writing and Manuscript Verification". According to the selected process nodes, a target execution sub-process is generated, and the target execution sub-process is "Analysis-Research-Material Organization-Article Writing-Manuscript Verification".

[0151] Step 406: Determine the target process node included in the target execution sub-process and the node identifier corresponding to the target process node.

[0152] In a specific embodiment of the present specification, a target process node is determined in a target execution sub-process, and a node identifier of the target process node is determined.

[0153] Step 408: Select a target functional interface corresponding to the target process node from the preset interface set corresponding to the target domain according to the node identifier, and use the target functional interface as the functional interface corresponding to the text generation task.

[0154] In a specific embodiment of the present specification, a target functional interface having an association relationship with a target process node is selected from a preset interface set according to a node identifier, and the determined target functional interfaces are all used as functional interfaces for a text generation task, including an analysis interface, a search interface, an organization interface, a generation interface, and a proofreading interface.

[0155] Step 410: Call the functional interface to execute the text generation task according to the target execution sub-process to obtain the target text corresponding to the text generation task.

[0156] In a specific embodiment of this specification, an AI agent uses an analysis interface to analyze task execution information and obtain corresponding topic information and text generation information. The topic information includes the user-entered topic "Hot Event A," and the text generation information includes user-specified creative requirements, such as article length and style. The topic information and text generation information are then output to a search interface.

[0157] The search interface is called to search for material data based on the subject information, and the material data associated with the subject information is obtained. The AI ​​agent automatically determines whether a multi-step search is required based on the material data obtained from the search, determines that the execution strategy of the search interface is a continued execution strategy, calls the search interface to generate search keywords based on the material data obtained from the search, and continues the search process based on the search keywords until the search stop condition is met. The material data obtained from the search is output to the organization interface.

[0158] The organization interface is called to aggregate and process the material data according to preset organization rules, including material recall and sorting, outline generation, and event excerpting. After aggregation, text-related information is obtained, including outline, summary, and description text. This text-related information is output to the generation interface.

[0159] The generation interface is called to perform text generation processing based on the text association information to obtain the initial text corresponding to the text generation task. The initial text is the draft of the news associated with "Hot Event A". The initial text can be proofread and revised later to obtain the final version of the press release. Therefore, the generation interface outputs the generated initial text to the proofreading interface.

[0160] The proofreading interface is called to adjust the initial text based on the text generation information according to the preset proofreading rules to obtain the target text corresponding to the text generation task. The target text is the final news draft that is finally fed back to the user.

[0161] This specification provides a text generation method that realizes the determination of the text generation task and the task execution information corresponding to the text generation task through an intelligent decision-making component, and determines the target execution sub-process in the preset execution process based on the task execution information, so that the intelligent decision-making component automatically decides to complete the execution process of the text generation task, reduces labor costs, and improves text generation efficiency. Subsequently, according to the target execution sub-process, the functional interface corresponding to the text generation task is selected from the preset interface set corresponding to the target field, and the functional interface is called to execute the text generation task according to the target execution sub-process to obtain the target text corresponding to the text generation task. By selecting the functional interface corresponding to the text generation task according to the target execution sub-process and calling the functional interface to complete the text generation task, the cost overhead caused by manual processing is further reduced, and the complexity of text generation is reduced.

[0162] See also Figure 5 , Figure 5 A flowchart of another text generation method provided according to an embodiment of this specification is shown. The method is applied to an intelligent decision-making component and specifically includes the following steps.

[0163] Step 502: Determine a text generation task and task execution information corresponding to the text generation task, wherein the task execution information is used to determine the task requirements of the text generation task.

[0164] Step 504: Determine a target execution sub-process in a preset execution process according to the task execution information.

[0165] Step 506: Execute a sub-process according to the target and select a functional interface corresponding to the text generation task.

[0166] Step 508: calling the functional interface to execute the text generation task according to the target execution sub-process, and obtaining the target news text corresponding to the text generation task.

[0167] In a specific embodiment of the present specification, the text generation method is applied in the news field to generate press releases related to events input by users. In order not to repeat the same contents of the technical solution, and the technical solution of the text generation belongs to the same concept as the technical solution of the above-mentioned text generation method, the details that are not described in detail can be found in the description of the technical solution of the above-mentioned text generation method.

[0168] This specification provides a text generation method, which is applied to an intelligent decision-making component, including determining a text generation task and task execution information corresponding to the text generation task, wherein the task execution information is used to determine the task requirements of the text generation task; determining a target execution sub-process in a preset execution process based on the task execution information; selecting a functional interface corresponding to the text generation task based on the target execution sub-process; calling the functional interface to execute the text generation task according to the target execution sub-process, and obtaining the target news text corresponding to the text generation task. The method realizes determining the text generation task and the task execution information corresponding to the text generation task by the intelligent decision-making component, determining the target execution sub-process in the preset execution process based on the task execution information, so that the intelligent decision-making component automatically decides to complete the execution process of the text generation task, reduces labor costs, and improves text generation efficiency. Subsequently, according to the target execution sub-process, the functional interface corresponding to the text generation task is selected from the preset interface set corresponding to the news field, and the functional interface is called to execute the text generation task according to the target execution sub-process, and the target news text corresponding to the text generation task is obtained. By selecting the functional interface corresponding to the text generation task according to the target execution sub-process and calling the functional interface to complete the text generation task, the cost overhead caused by manual processing is further reduced and the complexity of news text generation is reduced.

[0169] See also Figure 6 , Figure 6 A flowchart of another text generation method provided according to an embodiment of this specification is shown. The method is applied to an intelligent decision-making component and specifically includes the following steps.

[0170] Step 602: Receive a text generation task, and determine task execution information corresponding to the text generation task, wherein the task execution information is used to determine task requirements of the text generation task.

[0171] Step 604: Input the task execution information into the text generation model to obtain the target text corresponding to the text generation task output by the text generation model based on the task execution information, wherein the target text is generated by the text generation model according to the above-mentioned text generation method.

[0172] In a specific embodiment of the present specification, automatic text generation can be achieved through a text generation model. The text generation model can be understood as a model that executes the above-mentioned text generation method, including calling intelligent decision-making components and other functional interfaces, so that after the task execution information is input into the text generation model, the target text output by the text generation model can be obtained.

[0173] This specification provides a text generation method, which is applied to an intelligent decision-making component, receives a text generation task, determines task execution information corresponding to the text generation task, wherein the task execution information is used to determine the task requirements of the text generation task; inputs the task execution information into a text generation model, and obtains a target text corresponding to the text generation task output by the text generation model based on the task execution information, wherein the target text is generated by the text generation model according to the above-mentioned text generation method. The intelligent decision-making component determines the text generation task and the task execution information corresponding to the text generation task, and determines the target execution sub-process in a preset execution process based on the task execution information, so that the intelligent decision-making component automatically decides to complete the execution process of the text generation task, reducing labor costs and improving text generation efficiency. Subsequently, according to the target execution sub-process, the functional interface corresponding to the text generation task is selected from the preset interface set corresponding to the news field, and the functional interface is called to execute the text generation task according to the target execution sub-process, thereby obtaining the target news text corresponding to the text generation task. By selecting the functional interface corresponding to the text generation task according to the target execution sub-process and calling the functional interface to complete the text generation task, the cost overhead caused by manual processing is further reduced, and the complexity of news text generation is reduced.

[0174] See also Figure 7 , Figure 7 A flowchart of another text generation method provided according to an embodiment of this specification is shown. The method is applied to a cloud-side device and specifically includes the following steps.

[0175] Step 702: Receive the text generation task sent by the terminal-side device and determine task execution information corresponding to the text generation task, wherein the task execution information is used to determine the task requirements of the text generation task.

[0176] Step 704: Input the task execution information into the text generation model, obtain the target text corresponding to the text generation task output by the text generation model based on the task execution information, and send the target text to the terminal device, wherein the target text is generated by the text generation model according to the above-mentioned text generation method.

[0177] In a specific embodiment of this specification, a cloud-side device can be understood as a cloud computing device that provides text generation services, and an end-side device can be understood as a terminal device that users use to access text generation services. A service provider can receive text generation tasks from the end-side device, automatically generate text, and provide feedback to the user. A text generation model can be understood as a model that executes the aforementioned text generation method, including invoking intelligent decision-making components and other functional interfaces, so that after inputting task execution information into the text generation model, the target text output by the text generation model can be obtained.

[0178] Optionally, the method further includes receiving an adjustment instruction returned by the terminal-side device for the target text, determining adjustment information corresponding to the adjustment instruction; adjusting the target text based on the adjustment information, and returning the adjusted target text to the terminal-side device.

[0179] In a specific embodiment of the present specification, an adjustment instruction can be understood as an instruction for modifying the target text proposed by the user. For example, if the user is not satisfied with the currently generated target text, the adjustment instruction can be an instruction to regenerate the current target text. Alternatively, if the user is dissatisfied with certain content in the target text or has filtered out erroneous parts, an adjustment instruction can be issued for these parts. After receiving the adjustment instruction, adjustment information can be determined. The adjustment information includes the information that needs to be adjusted input by the user, such as adjustments to the entire target text or adjustments to parts of the target text. Based on the adjustment information, the target text can be adjusted and the adjusted target text is returned to the user so that the user obtains a text that meets the expected requirements.

[0180] The present specification provides a text generation method, which is applied to a cloud-side device, including receiving a text generation task sent by an end-side device, determining the task execution information corresponding to the text generation task, wherein the task execution information is used to determine the task requirements of the text generation task; inputting the task execution information into a text generation model, obtaining the target text corresponding to the text generation task output by the text generation model based on the task execution information, and sending the target text to the end-side device, wherein the target text is generated by the text generation model according to the above-mentioned text generation method. It realizes determining the text generation task and the task execution information corresponding to the text generation task through an intelligent decision-making component, determining the target execution sub-process in the preset execution process according to the task execution information, so that the intelligent decision-making component automatically decides to complete the execution process of the text generation task, reduces labor costs, and improves text generation efficiency. Subsequently, according to the target execution sub-process, the functional interface corresponding to the text generation task is selected from the preset interface set corresponding to the target field, and the functional interface is called to execute the text generation task according to the target execution sub-process to obtain the target text corresponding to the text generation task. By selecting the functional interface corresponding to the text generation task according to the target execution sub-process and calling the functional interface to complete the text generation task, the cost overhead caused by manual processing is further reduced and the complexity of text generation is reduced.

[0181] Corresponding to the above method embodiment, this specification also provides a text generation device embodiment, Figure 8 FIG. 1 shows a schematic diagram of a text generation device provided by an embodiment of this specification. Figure 8 As shown, the device includes:

[0182] A first determining module 802 is configured to determine a text generation task and task execution information corresponding to the text generation task, wherein the task execution information is used to determine a task requirement of the text generation task;

[0183] The second determining module 804 is configured to determine a target execution sub-process in a preset execution process according to the task execution information;

[0184] A selection module 806 is configured to select a functional interface corresponding to the text generation task from a preset interface set corresponding to the target domain according to the target execution sub-process;

[0185] The calling module 808 is configured to call the functional interface to execute the text generation task according to the target execution sub-process, and obtain the target text corresponding to the text generation task.

[0186] Optionally, the second determination module 804 is further configured to determine the task target corresponding to the task execution information, select a process node corresponding to the task target in a preset execution process, and generate a target execution sub-process based on the process node.

[0187] Optionally, the selection module 806 is further configured to determine the target process node included in the target execution sub-process, and the node identifier corresponding to the target process node; select the target functional interface corresponding to the target process node from the preset interface set corresponding to the target field according to the node identifier; and use the target functional interface as the functional interface corresponding to the text generation task.

[0188] Optionally, the selection module 806 is further configured so that the functional interface includes at least one of an analysis interface, a search interface, an organization interface, a generation interface, a proofreading interface and an auxiliary interface.

[0189] Optionally, the calling module 808 is further configured to call the analysis interface to analyze the task execution information to obtain text generation information corresponding to the text generation task; and output the text generation information according to the target execution sub-process.

[0190] Optionally, the calling module 808 is further configured to call the search interface to perform search processing based on the text generation information to obtain material data associated with the text generation information; and output the material data according to the target execution sub-process.

[0191] Optionally, the calling module 808 is further configured to determine the execution strategy of the search interface based on the material data; when the execution strategy is a continue execution strategy, the search interface is called to generate search keywords based on the material data, the search keywords are used as the subject information, and the search interface is called to continue to perform the step of searching based on the subject information; when the execution strategy is a stop execution strategy, the search interface is called to continue to perform the step of outputting the material data according to the target execution sub-process.

[0192] Optionally, the calling module 808 is further configured to call the organization interface to summarize and process the material data according to preset organization rules to obtain target material data in the material data, wherein the preset organization rules are used to determine the organization conditions for the material data; generate text association information based on the target material data, and output the text association information according to the target execution sub-process.

[0193] Optionally, the calling module 808 is further configured to call the generation interface to perform text generation processing according to the text association information to obtain the initial text corresponding to the text generation task; and output the initial text according to the target execution sub-process.

[0194] Optionally, the calling module 808 is further configured to call the proofreading interface to adjust the initial text based on the text generation information according to preset proofreading rules to obtain the target text corresponding to the text generation task, wherein the preset proofreading rules are used to determine the proofreading conditions for the text generation information; and output the target text according to the target execution sub-process.

[0195] Optionally, the calling module 808 is further configured to call the auxiliary interface to extract information from the target text to obtain auxiliary text associated with the target text; and output the auxiliary text according to the target execution sub-process.

[0196] This specification provides a text generation device that determines the text generation task and the task execution information corresponding to the text generation task through an intelligent decision-making component, and determines the target execution sub-process in the preset execution process based on the task execution information, so that the intelligent decision-making component automatically decides to complete the execution process of the text generation task, reducing labor costs and improving text generation efficiency. Subsequently, according to the target execution sub-process, the functional interface corresponding to the text generation task is selected from the preset interface set corresponding to the target field, and the functional interface is called to execute the text generation task according to the target execution sub-process to obtain the target text corresponding to the text generation task. By selecting the functional interface corresponding to the text generation task according to the target execution sub-process and calling the functional interface to complete the text generation task, the cost overhead caused by manual processing is further reduced, and the complexity of text generation is reduced.

[0197] The above is a schematic diagram of a text generation device according to this embodiment. It should be noted that the technical solution of the text generation device and the technical solution of the above-mentioned text generation method are based on the same concept. For details not described in detail in the technical solution of the text generation device, please refer to the description of the technical solution of the above-mentioned text generation method.

[0198] Corresponding to the above method embodiment, this specification also provides a text generation device embodiment, Figure 9 FIG. 1 shows a schematic diagram of the structure of another text generation device provided by an embodiment of this specification. Figure 9 As shown, the device includes:

[0199] A first determining module 902 is configured to determine a text generation task and task execution information corresponding to the text generation task, wherein the task execution information is used to determine a task requirement of the text generation task;

[0200] The second determining module 904 is configured to determine a target execution sub-process in a preset execution process according to the task execution information;

[0201] A selection module 906 is configured to execute a sub-process according to the target and select a functional interface corresponding to the text generation task;

[0202] The calling module 908 is configured to call the functional interface to execute the text generation task according to the target execution sub-process, and obtain the target news text corresponding to the text generation task.

[0203] This specification provides a text generation device that determines the text generation task and the task execution information corresponding to the text generation task through an intelligent decision-making component, and determines the target execution sub-process in the preset execution process based on the task execution information, so that the intelligent decision-making component automatically decides to complete the execution process of the text generation task, reducing labor costs and improving text generation efficiency. Subsequently, according to the preset interface set corresponding to the target execution sub-process in the news field, the functional interface corresponding to the text generation task is selected, and the functional interface is called to execute the text generation task according to the target execution sub-process to obtain the target news text corresponding to the text generation task. By selecting the functional interface corresponding to the text generation task according to the target execution sub-process and calling the functional interface to complete the text generation task, the cost overhead caused by manual processing is further reduced, and the complexity of news text generation is reduced.

[0204] The above is a schematic diagram of a text generation device according to this embodiment. It should be noted that the technical solution of the text generation device and the technical solution of the above-mentioned text generation method are based on the same concept. For details not described in detail in the technical solution of the text generation device, please refer to the description of the technical solution of the above-mentioned text generation method.

[0205] Corresponding to the above method embodiment, this specification also provides a text generation device embodiment, Figure 10 FIG. 1 shows a schematic diagram of the structure of another text generation device provided by an embodiment of this specification. Figure 10 As shown, the device includes:

[0206] The receiving module 1002 is configured to receive a text generation task and determine task execution information corresponding to the text generation task, wherein the task execution information is used to determine task requirements of the text generation task;

[0207] The input module 1004 is configured to input the task execution information into the text generation model to obtain the target text corresponding to the text generation task output by the text generation model based on the task execution information, wherein the target text is generated by the text generation model according to the above-mentioned text generation method.

[0208] This specification provides a text generation device that determines the text generation task and the task execution information corresponding to the text generation task through an intelligent decision-making component, and determines the target execution sub-process in the preset execution process based on the task execution information, so that the intelligent decision-making component automatically decides to complete the execution process of the text generation task, reducing labor costs and improving text generation efficiency. Subsequently, according to the target execution sub-process, the functional interface corresponding to the text generation task is selected from the preset interface set corresponding to the target field, and the functional interface is called to execute the text generation task according to the target execution sub-process to obtain the target text corresponding to the text generation task. By selecting the functional interface corresponding to the text generation task according to the target execution sub-process and calling the functional interface to complete the text generation task, the cost overhead caused by manual processing is further reduced, and the complexity of text generation is reduced.

[0209] The above is a schematic diagram of a text generation device according to this embodiment. It should be noted that the technical solution of the text generation device and the technical solution of the above-mentioned text generation method are based on the same concept. For details not described in detail in the technical solution of the text generation device, please refer to the description of the technical solution of the above-mentioned text generation method.

[0210] Corresponding to the above method embodiment, this specification also provides a text generation device embodiment, Figure 11 FIG. 1 shows a schematic diagram of the structure of another text generation device provided by an embodiment of this specification. Figure 11 As shown, the device includes:

[0211] The receiving module 1102 is configured to receive a text generation task sent by a terminal-side device and determine task execution information corresponding to the text generation task, wherein the task execution information is used to determine a task requirement of the text generation task;

[0212] The sending module 1104 is configured to input the task execution information into the text generation model, obtain the target text corresponding to the text generation task output by the text generation model based on the task execution information, and send the target text to the terminal device, wherein the target text is generated by the text generation model according to the above-mentioned text generation method.

[0213] Optionally, the device also includes an adjustment module, which is configured to receive an adjustment instruction returned by the terminal-side device for the target text, determine adjustment information corresponding to the adjustment instruction; adjust the target text based on the adjustment information, and return the adjusted target text to the terminal-side device.

[0214] This specification provides a text generation device that receives a text generation task sent by a terminal device, determines the corresponding task execution information, and outputs the target text corresponding to the text generation task based on the task execution information based on a text generation model. This device implements the text generation task through the text generation model, reducing labor costs and improving text generation efficiency. The intelligent decision-making component in the text generation model automatically determines the execution process for completing the text generation task and calls a functional interface to complete the text generation task, reducing the cost overhead caused by manual processing and the complexity of text generation, thereby providing a convenient and fast text generation service for terminal devices.

[0215] The above is a schematic diagram of a text generation device according to this embodiment. It should be noted that the technical solution of the text generation device and the technical solution of the above-mentioned text generation method are based on the same concept. For details not described in detail in the technical solution of the text generation device, please refer to the description of the technical solution of the above-mentioned text generation method.

[0216] Figure 12 The following is a block diagram of a computing device 1200 according to one embodiment of the present disclosure. Components of the computing device 1200 include, but are not limited to, a memory 1210 and a processor 1220. The processor 1220 is connected to the memory 1210 via a bus 1230, and a database 1250 is used to store data.

[0217] The computing device 1200 also includes an access device 1240 that enables the computing device 1200 to communicate via one or more networks 1260. Examples of such networks include a public switched telephone network (PSTN), a local area network (LAN), a wide area network (WAN), a personal area network (PAN), or a combination of communication networks such as the Internet. The access device 1240 may include one or more of any type of network interface (e.g., a network interface card (NIC)) whether wired or wireless, such as an IEEE 802.11 wireless local area network (WLAN) wireless interface, a Worldwide Interoperability for Microwave Access (Wi-MAX) interface, an Ethernet interface, a universal serial bus (USB) interface, a cellular network interface, a Bluetooth interface, or a near field communication (NFC) interface.

[0218] In one embodiment of the present specification, the above components of the computing device 1200 and Figure 12 Other components not shown in the figure may also be connected to each other, for example, via a bus. Figure 12 The computing device structure block diagram shown is for illustrative purposes only and is not intended to limit the scope of this specification. Those skilled in the art may add or replace other components as needed.

[0219] Computing device 1200 may be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., a tablet computer, personal digital assistant, laptop computer, notebook computer, netbook computer, etc.), a mobile phone (e.g., a smartphone), a wearable computing device (e.g., a smartwatch, smart glasses, etc.), or other types of mobile devices, or a stationary computing device such as a desktop computer or personal computer (PC). Computing device 1200 may also be a mobile or stationary server.

[0220] The processor 1220 is configured to execute the following computer-executable instructions, which, when executed by the processor, implement the steps of the above-mentioned text generation method.

[0221] The above is a schematic diagram of a computing device according to this embodiment. It should be noted that the technical solution of the computing device and the technical solution of the above-mentioned text generation method are based on the same concept. For details not described in detail in the technical solution of the computing device, please refer to the description of the technical solution of the above-mentioned text generation method.

[0222] An embodiment of the present specification further provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the above-mentioned text generation method.

[0223] The above is a schematic diagram of a computer-readable storage medium according to this embodiment. It should be noted that the technical solution of this storage medium and the technical solution of the above-mentioned text generation method are based on the same concept. For details not described in detail in the technical solution of the storage medium, please refer to the description of the technical solution of the above-mentioned text generation method.

[0224] An embodiment of the present specification further provides a computer program product, including a computer program or instructions, which implements the steps of the above-mentioned text generation method when executed by a processor.

[0225] The above is a schematic diagram of a computer program product according to this embodiment. It should be noted that the technical solution of this computer program product and the technical solution of the aforementioned text generation method are based on the same concept. For details not described in detail in the technical solution of the computer program product, please refer to the description of the technical solution of the aforementioned text generation method.

[0226] The foregoing description of this specification describes specific embodiments. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or the sequential order to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0227] The computer instructions include computer program code, which may be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal, and software distribution medium. It should be noted that the content contained in the computer-readable medium may be appropriately increased or decreased according to the requirements of patent practice. For example, in some regions, according to patent practice, computer-readable media does not include electric carrier signals and telecommunication signals.

[0228] It should be noted that for the aforementioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the embodiments of this specification are not limited by the order of the actions described, because according to the embodiments of this specification, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the embodiments of this specification.

[0229] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0230] The preferred embodiments disclosed above are intended only to help illustrate this specification. The optional embodiments do not exhaustively describe all details, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations can be made based on the content of the embodiments of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the embodiments of this specification, so that those skilled in the art can better understand and utilize this specification. This specification is limited only by the claims and their full scope and equivalents.

Claims

1. A text generation method, applied to an intelligent decision-making component, comprising: Determining a text generation task and task execution information corresponding to the text generation task, wherein the task execution information is used to determine task requirements of the text generation task; Determining a target execution sub-process in a preset execution process based on the task execution information; Execute the sub-process according to the target and select the functional interface corresponding to the text generation task; The functional interface is called to execute the text generation task according to the target execution sub-process to obtain the target text corresponding to the text generation task.

2. The method according to claim 1, wherein determining a target execution sub-process in a preset execution process according to the task execution information comprises: Determining a task objective corresponding to the task execution information; Select a process node corresponding to the task target in the preset execution process; A target execution sub-process is generated based on the process node.

3. The method according to claim 1, wherein the step of selecting a functional interface corresponding to the text generation task according to the target execution sub-process comprises: Determine a target process node included in the target execution sub-process and a node identifier corresponding to the target process node; Selecting a target functional interface corresponding to the target process node from a preset interface set corresponding to the target domain according to the node identifier; The target functional interface is used as the functional interface corresponding to the text generation task. 4 . The method according to claim 1 , wherein the functional interface comprises at least one of an analysis interface, a search interface, an organization interface, a generation interface, a proofreading interface, and an auxiliary interface.

5. The method according to claim 4, wherein the functional interface is the analysis interface; in, Calling the functional interface to execute the text generation task according to the target execution sub-process includes: Calling the analysis interface to analyze the task execution information to obtain text generation information corresponding to the text generation task; The sub-process is executed according to the target to output the text generation information.

6. The method according to claim 5, wherein the functional interface is the search interface; in, Calling the functional interface to execute the text generation task according to the target execution sub-process includes: Calling the search interface to perform search processing based on the text generation information to obtain material data associated with the text generation information; The material data is outputted according to the target execution sub-process.

7. The method according to claim 6, after obtaining the material data associated with the text generation information, further comprising: determining an execution strategy of the search interface according to the material data; In the case where the execution strategy is the continue execution strategy, calling the search interface to generate a search keyword based on the material data, using the search keyword as the text generation information, and calling the search interface to continue to perform the step of performing a search process based on the text generation information; In a case where the execution strategy is a stop execution strategy, the search interface is called to continue the step of outputting the material data according to the target execution sub-process.

8. The method according to claim 6, wherein the functional interface is the tissue interface; in, Calling the functional interface to execute the text generation task according to the target execution sub-process includes: calling the organization interface to aggregate the material data according to a preset organization rule to obtain target material data from the material data, wherein the preset organization rule is used to determine an organization condition for the material data; Generating text association information according to the target material data, and outputting the text association information according to the target execution sub-process.

9. The method according to claim 8, wherein the functional interface is the generation interface; in, Calling the functional interface to execute the text generation task according to the target execution sub-process includes: Calling the generation interface to perform text generation processing according to the text association information to obtain the initial text corresponding to the text generation task; The sub-process is executed according to the target to output the initial text.

10. The method according to claim 9, wherein the functional interface is the proofreading interface; in, Calling the functional interface to execute the text generation task according to the target execution sub-process includes: Calling the proofreading interface to adjust the initial text based on the text generation information according to preset proofreading rules to obtain a target text corresponding to the text generation task, wherein the preset proofreading rules are used to determine proofreading conditions for the text generation information; The target text is output according to the target execution sub-process.

11. The method according to claim 10, wherein the functional interface is the auxiliary interface; in, Calling the functional interface to execute the text generation task according to the target execution sub-process includes: Calling the auxiliary interface to extract information from the target text to obtain auxiliary text associated with the target text; The auxiliary text is outputted according to the target execution sub-process.

12. A text generation method, applied to an intelligent decision-making component, comprising: Determining a text generation task and task execution information corresponding to the text generation task, wherein the task execution information is used to determine task requirements of the text generation task; Determining a target execution sub-process in a preset execution process based on the task execution information; Execute the sub-process according to the target and select the functional interface corresponding to the text generation task; The functional interface is called to execute the text generation task according to the target execution sub-process to obtain the target news text corresponding to the text generation task.

13. A text generation method, applied to an intelligent decision-making component, comprising: Receive a text generation task, and determine task execution information corresponding to the text generation task, wherein the task execution information is used to determine task requirements of the text generation task; The task execution information is input into a text generation model to obtain a target text corresponding to the text generation task output by the text generation model based on the task execution information, wherein the target text is generated by the text generation model according to the method described in any one of claims 1 to 12.

14. A text generation method, applied to a cloud-side device, comprising: receiving a text generation task sent by a terminal-side device, and determining task execution information corresponding to the text generation task, wherein the task execution information is used to determine a task requirement of the text generation task; The task execution information is input into a text generation model to obtain a target text corresponding to the text generation task output by the text generation model based on the task execution information, and the target text is sent to the terminal device, wherein the target text is generated by the text generation model according to the method described in any one of claims 1 to 12.

15. The method according to claim 14, further comprising: receiving an adjustment instruction returned by the terminal device for the target text, and determining adjustment information corresponding to the adjustment instruction; The target text is adjusted based on the adjustment information, and the adjusted target text is returned to the terminal-side device.

16. A text generation system, comprising a client and a server, wherein: The client is used to send a text generation task to the server; The server is used to determine the task execution information corresponding to the text generation task, wherein the task execution information is used to determine the task requirements of the text generation task, determine the target execution sub-process in the preset execution process according to the task execution information, select the functional interface corresponding to the text generation task according to the target execution sub-process, call the functional interface to execute the text generation task according to the target execution sub-process, obtain the target text corresponding to the text generation task, and send the target text to the client.

17. A computing device comprising: memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the method according to any one of claims 1 to 15 are implemented.

18. A computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions, when executed by a processor, implement the steps of the method according to any one of claims 1 to 15.

19. A computer program product comprising a computer program or instructions, which implement the steps of the method according to any one of claims 1 to 15 when executed by a processor.

Images