Prompt word generation method, text processing method, electronic equipment and storage medium

Abstract

The invention discloses a cue word generation method, a text processing method, electronic equipment and a storage medium. The method comprises the following steps: in response to a first input instruction acting on an operation interface, displaying first text data of a natural language on the operation interface; a conversion instruction acting on the operation interface is responded, a conversion window is displayed on the operation interface, second text data of the target language are displayed in the conversion window, and the second text data are text data obtained by conducting language conversion on the first text data; a first publishing instruction acting on the operation interface is responded, a first publishing result corresponding to the first publishing instruction is displayed on the operation interface, and the first publishing result is used for representing a result of generating the target prompt word based on the second text data. The technical problem that cue word generation is low in accuracy and efficiency in the prior art is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of large model and intelligent customer service, in particular to a prompt word generation method, a text processing method, an electronic device and a storage medium. BACKGROUND

[0002] In the field of artificial intelligence, entities with artificial intelligence capabilities, i.e., agents, can perform specific tasks or provide services. Agents can be applied in robots to make robots more intelligent, enabling them to perform tasks autonomously and interact with humans. Agents can also be used to develop chat robots for natural language communication with users, providing information queries, customer service support, and other services. Generating prompt words for agents can help users better understand and use agents, improve user experience, guide users to perform correct operations, answer common questions, provide help and suggestions, and enhance user interaction with agents. It is an important step in the development of agents.

[0003] Currently, the generation of prompt words in related technologies usually needs to be based on pre-arranged and developed processes. The generation process of prompt words is relatively complex and time-consuming. Users need to be familiar with the development platform and programming language related to prompt word development, which results in high learning cost and programming time cost for users to generate prompt words. Moreover, the generated prompt words often cannot be adjusted in a timely and convenient manner according to user needs, and the generated prompt words are difficult to meet user needs, resulting in low accuracy and efficiency of generating prompt words.

[0004] To address the above problems, no effective solutions have been proposed so far. SUMMARY

[0005] Embodiments of the present application provide a prompt word generation method, a text processing method, an electronic device and a storage medium to at least solve the technical problem of low accuracy and efficiency of generating prompt words in related technologies.

[0006] According to an aspect of the embodiments of the present application, a method for generating a prompt word is provided. The method comprises: in response to a first input instruction acting on an operation interface, displaying first text data in a natural language on the operation interface, wherein the first text data is used to describe an execution process of a target task performed by a question and answer processing model; in response to a conversion instruction acting on the operation interface, displaying a conversion window on the operation interface, wherein second text data in a target language is displayed in the conversion window, the second text data being text data obtained by language conversion on the first text data, and the target language being used to represent a language corresponding to a field to which the target task belongs; and in response to a first publishing instruction acting on the operation interface, displaying a first publishing result corresponding to the first publishing instruction on the operation interface, wherein the first publishing result is used to represent a result of generating a target prompt word based on the second text data, and the target prompt word is used to guide the question and answer processing model to perform the target task.

[0007] According to another aspect of the embodiments of the present application, a text processing method is also provided. The method comprises: in response to an input instruction acting on an interactive interface, displaying a current inquiry text on the interactive interface; and in response to a processing instruction acting on the interactive interface, displaying a current reply text corresponding to the current inquiry text on the interactive interface, wherein the current reply text is obtained by processing the current inquiry text by a question and answer processing model guided by a target prompt word, and the target prompt word is a prompt word generated by the above method.

[0008] According to another aspect of the embodiments of the present application, a method for generating a prompt word is also provided. The method comprises: obtaining first text data in a natural language, wherein the first text data is used to describe an execution process of a target task performed by a question and answer processing model; performing language conversion on the first text data to obtain second text data in a target language, wherein the target language is used to represent a language corresponding to a field to which the target task belongs; and generating a target prompt word based on the second text data, wherein the target prompt word is used to guide the question and answer processing model to perform the target task.

[0009] According to another aspect of the embodiments of the present application, a method for generating a prompt word is also provided. The method comprises: obtaining first text data in a natural language by calling a first interface, wherein the first interface comprises a first parameter, a parameter value of the first parameter comprises the first text data, and the first text data is used to describe an execution process of a target task performed by a question and answer processing model; performing language conversion on the first text data to obtain second text data in a target language, wherein the target language is used to represent a language corresponding to a field to which the target task belongs; generating a target prompt word based on the second text data, wherein the target prompt word is used to guide the question and answer processing model to perform the target task; and outputting the target prompt word by calling a second interface, wherein the second interface comprises a second parameter, and a parameter value of the second parameter comprises the target prompt word.

[0010] According to another aspect of the embodiments of this application, an electronic device is also provided, including: a memory storing an executable program; and a processor for running the program, wherein the program executes the methods in various embodiments of this application when it runs.

[0011] According to another aspect of the embodiments of this application, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the computer-readable storage medium is located to perform the methods of various embodiments of this application.

[0012] According to another aspect of the embodiments of this application, a computer program product is also provided, including a computer program that, when executed by a processor, implements the methods of various embodiments of this application.

[0013] According to another aspect of the embodiments of this application, a computer program product is also provided, including a non-volatile computer-readable storage medium storing a computer program that, when executed by a processor, implements the methods in various embodiments of this application.

[0014] According to another aspect of the embodiments of this application, a computer program is also provided, which, when executed by a processor, implements the methods of the various embodiments of this application.

[0015] The solution provided by the above embodiments of this application, in response to a first input command applied to the operation interface, displays first text data in natural language on the operation interface, wherein the first text data is used to describe the execution process of the target task performed by the question-answering processing model; in response to a conversion command applied to the operation interface, displays a conversion window on the operation interface, wherein the conversion window displays second text data in the target language, the second text data being text data obtained by language conversion of the first text data, and the target language being used to represent the language corresponding to the domain to which the target task belongs; in response to a first release command applied to the operation interface, displays a first release result corresponding to the first release command on the operation interface, wherein the first release result is used to represent the result of generating target prompt words based on the second text data, and the target prompt words are used to guide the question-answering processing model to perform the target task. It is noteworthy that the system can automatically generate prompt words by responding to user input commands, conversion commands, and release commands. Users can input natural language to describe the target task of the question-answering processing model, and the system can display the task execution process based on the user's input text data to help users confirm task requirements. Users can use conversion commands to convert the displayed text data into text data in the target language to obtain the task requirements and determine the linguistic context of the required prompt words. The system can respond to the user's first release command to generate and display prompt word results based on the target language text data, guiding the question-answering processing model to execute the target task. Specifically, based on the conversion window, users can understand the tasks being performed by the system in real time and intuitively, clearly see the converted second text data, facilitate the determination of the correctness of the second text data, and make timely adjustments or corrections to the second text data. The generated prompt words can meet user needs and achieve accurate generation of prompt words. Users can directly describe the target task based on their familiar natural language and interact with the system's intuitive and user-friendly interface through simple command operations. Users do not need to understand or master the complex programming languages ​​that the question-answering processing model can handle. Users can only focus on the logic of the target task itself, reducing the learning cost and programming time cost of generating target prompt words, achieving efficient generation of prompt words, and thus solving the technical problems of low accuracy and efficiency in generating prompt words in related technologies.

[0016] It is worth noting that the general description above and the detailed description that follow are merely for illustrative purposes and do not constitute a limitation on this application. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0018] Figure 1This is a schematic diagram illustrating an application scenario of a prompt word generation method according to an embodiment of this application;

[0019] Figure 2 This is a flowchart of a method for generating prompt words according to an embodiment of this application;

[0020] Figure 3 This is a schematic diagram of an optional target prompt word generation process according to an embodiment of this application;

[0021] Figure 4 This is a schematic diagram of another optional target prompt word generation process according to an embodiment of this application;

[0022] Figure 5 This is a schematic diagram of an optional user interface according to an embodiment of this application;

[0023] Figure 6 This is a schematic diagram of an optional debugging window according to an embodiment of this application;

[0024] Figure 7 This is a schematic diagram of an optional response to a viewing command applied to a debug window according to an embodiment of this application;

[0025] Figure 8 This is a flowchart of a text processing method according to an embodiment of this application;

[0026] Figure 9 This is a flowchart of an optional prompt word generation method according to an embodiment of this application;

[0027] Figure 10 This is a flowchart of another optional method for generating prompt words according to an embodiment of this application;

[0028] Figure 11 This is a schematic diagram of a prompt word generation device according to an embodiment of this application;

[0029] Figure 12 This is a schematic diagram of a text processing apparatus according to an embodiment of this application;

[0030] Figure 13 This is a schematic diagram of an optional prompt word generation device according to an embodiment of this application;

[0031] Figure 14 This is a schematic diagram of another optional prompt word generation device according to an embodiment of this application;

[0032] Figure 15 This is a structural block diagram of an electronic device according to an embodiment of this application. Detailed Implementation

[0033] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0034] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0035] The technical solution provided in this application is mainly implemented using large-scale model technology. Here, "large-scale model" refers to a deep learning model with a massive number of parameters, typically containing hundreds of millions, tens of billions, hundreds of billions, trillions, or even tens of trillions of parameters. Large-scale models can also be called foundation models. They are pre-trained using large-scale unlabeled corpora to produce pre-trained models with hundreds of millions of parameters. Such models can adapt to a wide range of downstream tasks and have good generalization ability. Examples include Large Language Models (LLMs) and multi-modal pre-training models.

[0036] It's important to note that in practical applications, large models can be fine-tuned using a small number of samples after pre-training, allowing them to be applied to various tasks. For example, large models can be widely used in Natural Language Processing (NLP), computer vision, and speech processing. Specifically, they can be applied to computer vision tasks such as Visual Question Answering (VQA), Image Captioning (IC), and Image Generation, as well as NLP tasks such as text-based sentiment classification, text summarization, and machine translation. Therefore, the main application scenarios for large models include, but are not limited to, digital assistants, intelligent robots, search, online education, office software, e-commerce, and intelligent design.

[0037] First, some nouns or terms that appear in the description of the embodiments of this application shall be interpreted as follows:

[0038] Service areas: This can refer to the support and assistance that a company or organization provides to customers through various channels, such as telephone, email, online chat, social media, etc., to ensure customer satisfaction, solve problems, answer questions, and provide consultation. Good customer service helps to build and maintain customer relationships, improve customer loyalty, and enhance brand reputation.

[0039] Large Language Models (LLMs) are artificial intelligence models based on deep learning technology. They are trained on large amounts of text data and can understand and generate natural language. They can perform a variety of tasks, such as text generation, translation, summarization, and question answering. Large language models typically have a large number of parameters, such as OpenAI's GPT-4 and Tongyi 1000 Questions, and can generate highly coherent and context-sensitive text.

[0040] Intelligent agent: In the field of artificial intelligence, an entity is capable of perceiving its environment and taking actions to achieve a specific goal. An intelligent agent can be a software program, a robot, or other computer system. Intelligent agents have the ability to learn and adapt, and can improve their behavior and decision-making through interaction with the environment.

[0041] Domain-Specific Language (DSL): This is a computer language designed specifically for a particular application domain or problem domain. DSLs are specifically designed in terms of syntax and functionality to meet the needs of a particular domain, providing higher abstraction and simplicity, thereby improving development efficiency and maintainability.

[0042] Prompt: In the field of natural language processing, a prompt is a piece of text used to guide or vaguely specify a task or input information. For large language models, a prompt can be a question, instruction, or contextual information used to trigger the model to generate a relevant and coherent text response.

[0043] Chain-of-Thought: This can be a reasoning technique and generative strategy used to improve the ability of large language models to understand and generate complex tasks. This method helps the model to plan and solve problems more clearly during the reasoning process by guiding the model to generate a series of step-by-step reasoning steps, i.e., chain-of-thought. It is suitable for problems that require multi-step logical reasoning, such as mathematical calculations, complex question answering, and logical inference.

[0044] According to an embodiment of this application, a method for generating prompt words is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0045] Considering the large number of model parameters in large models and the limited computing resources of mobile terminals, the prompt word generation method provided in this application embodiment can be applied to, for example, Figure 1 The application scenarios shown are not limited to these. In, for example... Figure 1 In the application scenario shown, the large model is deployed on server 10. Server 10 can connect to one or more client devices 20 via a local area network (LAN), wide area network (WAN), internet connection, or other types of data network. These client devices 20 may include, but are not limited to, smartphones, tablets, laptops, PDAs, personal computers, smart home devices, and in-vehicle devices. Client devices 20 can interact with users through a graphical user interface to access the large model, thereby implementing the method provided in this embodiment.

[0046] In this embodiment, the system consisting of a client device and a server can perform the following steps: The client device can acquire a first input instruction, a conversion instruction, and a first release instruction through a graphical user interface, display first text data in natural language on the operation interface, display a conversion window on the operation interface, and display the first release result corresponding to the first release instruction on the operation interface; the server can convert the first input instruction into first text data, perform language conversion on the first text data using a text conversion model to obtain second text data, and generate target prompt words based on the second text data to obtain the first release result. It should be noted that this embodiment can be performed on the client device if the client device's operating resources can meet the deployment and operation conditions of a large model. Under the above operating environment, this application provides the following... Figure 2 The method for generating the prompt words shown. Figure 2 This is a flowchart of a method for generating prompt words according to an embodiment of this application. For example... Figure 2 As shown, the specific steps may include the following:

[0047] Step S202: In response to the first input command applied to the operation interface, display the first text data in natural language on the operation interface.

[0048] The first text data describes the execution process of the target task performed by the question-answering processing model.

[0049] The aforementioned user interface can refer to a graphical user interface that allows users to interact with the system efficiently and conveniently, providing a good user experience. The user interface can be displayed on various mobile terminals, desktop computers, server-side operation / management systems, and other devices. The user interface can also be customized according to actual needs, which is not limited here.

[0050] The aforementioned first input instruction can refer to an instruction containing first text data that is input into the operation interface during the interaction between the user and the operation interface. The first input instruction can be an instruction generated by the user interacting with the operation interface through input devices such as keyboard, touch screen, button, and mouse, or it can refer to an instruction generated by the user interacting with the operation interface through voice, body movements, etc. The form of the first input instruction can also be determined according to actual needs, and is not limited here.

[0051] The aforementioned question-answering model can refer to an artificial intelligence model used to process natural language questions and provide corresponding responses / answers. The question-answering model can be based on deep learning technology and trained with the help of a large-scale corpus in order to understand and answer the questions raised by users. The question-answering model can be applied to various scenarios, such as intelligent assistants, online customer service, intelligent search engines, etc. The specific type of question-answering model can be determined according to actual needs, and is not limited here.

[0052] The aforementioned target task can refer to the task that the question-answering processing model needs to perform. It can be to understand the questions or needs raised by users and provide users with relevant information or solutions. The target task can include a series of sequential structures, judgment structures, or loop structures, etc.

[0053] The target task can include various application scenarios, such as intelligent customer service systems, intelligent search engines, and voice assistants. In the application scenario of intelligent customer service systems, users can ask questions or make requests through text or voice, and the system can understand the user's intent and provide corresponding answers or solutions. In the application scenario of intelligent search engines, users can enter search keywords, and the system can understand the user's search intent and return relevant search results. In the application scenario of voice assistants, users can ask questions or give commands through voice, and the system can understand the user's voice input and provide corresponding answers or perform corresponding operations. The target task can also be determined according to actual needs, and is not limited here.

[0054] The aforementioned first text data may refer to the execution process of the target task described by the user in natural language. The execution process of the target task described in natural language may include sequential structure, judgment structure, and loop structure. The form of the first text data may be written language, spoken language, dialect, or Braille and other body language of various countries or regions. The first text data may be determined according to the specific interaction between the user and the operation interface, and is not limited here.

[0055] In one optional embodiment, the system can respond to a first input command applied to the user interface via human-computer interaction. This can be an operation triggered by the user clicking a button, entering text, or similar actions. The user interface can display the first text data in natural language, allowing them to intuitively understand the task the system is performing and clearly identify the input text data. This facilitates the determination of the correctness of the first text data and timely correction. In particular, users can directly describe the target task based on their familiar natural language and interact with the system's intuitive and user-friendly interface. Users do not need to understand or master the complex programming languages ​​that the question-answering model can handle. Users can focus only on the logic of the target task itself, reducing the learning cost and programming time cost of generating target prompts and facilitating the subsequent generation of target prompts.

[0056] Step S204: In response to the conversion command applied to the operation interface, a conversion window is displayed on the operation interface.

[0057] The conversion window displays second text data in the target language. The second text data is obtained by converting the first text data into a different language. The target language is used to represent the language corresponding to the domain to which the target task belongs.

[0058] The aforementioned conversion instructions can refer to instructions generated by the user interacting with the interface through clicking, selecting, or using language, which are used to control the conversion of the first text language to the second text language. The form of the conversion instructions can be determined according to the specific interaction method between the user and the interface, and is not limited here.

[0059] The aforementioned conversion window can be a single window that dynamically displays the conversion of the first text language into the second text language, or it can be a dual window that compares and displays the first text language with the second text language. The form of the conversion window can be determined according to actual needs, and is not limited here.

[0060] The target language mentioned above can refer to the common language or syntax of the domain to which the question-answering processing model belongs, so as to facilitate the processing of the question-answering processing model. The target language can be determined according to the question-answering model and the domain to which the target task belongs. Here, there is no limitation on the target language.

[0061] Optionally, if the question-answering model is used to handle problems in the data analysis domain, the target language can be a Domain Specific Language (DSL). A DSL can contain specific keywords, syntax, and semantic rules to describe relevant information for the data analysis task. Alternatively, YAML (Yin't Markup Language) or eXtensible Markup Language (XML) can be used as the target language. YAML and XML can be used as configuration files or data exchange formats, and can well describe the input data or parameters required by the model. YAML or XML files can be used to specify the input of the question and the expected output, and then these files are passed to the question-answering model for processing. In addition, some question-answering models may also need a completely custom target language to describe problems in a specific domain. For example, a virtual assistant model may need a specific syntax to understand user instructions and perform corresponding operations. A custom language or syntax can be designed as the target language to help the model better understand the user's intent. The target language can be determined according to the domain to which the question-answering model and the target task belong. By designing a suitable target language, the performance of the question-answering model and the user experience can be improved.

[0062] The aforementioned second text data can refer to the first text data expressed in the target language. The second text data can semantically correspond to the first text language. For example, when the first text data is the target task that a question-and-answer processing model needs to process, described by a user in Chinese, and the target language is DSL, the second text data can be the first text data described in DSL. The second text data can be determined according to actual needs, and is not limited here.

[0063] In an optional embodiment, this application can use a text conversion model to convert the input first text data into second text data. The user can input the first text data through an interface and trigger the text conversion model to process it via a conversion command. A neural machine translation model can be used to perform language conversion on the first text data to obtain the second text data. The neural machine translation model can be a neural network-based machine translation method that learns the correspondence between languages ​​by training a large amount of parallel corpus, i.e., different language versions of the same text, thereby achieving text translation. In the neural machine translation model, an encoder-decoder structure can be used. The encoder receives the input text data and converts it into a fixed-length vector representation containing the semantic information of the input text. The decoder receives the vector representation output by the encoder and then generates text data in the target language, thus achieving language conversion of the first text data to obtain the second text data. Rule-based machine translation, statistical machine translation, and other methods can be used to perform language conversion on the first text data. The first text data is converted into the second text data through language conversion. Alternatively, the first text data can be converted into the second text data through manual translation, a specific language conversion algorithm, or a pre-defined mapping relationship between natural language and target language. The specific implementation method can be determined according to actual needs and is not limited here. The conversion command can control the display of the conversion window on the operation interface. The conversion window can display the second text data after language conversion. The target language belongs to the language corresponding to the domain of the target task, which can ensure that the generated prompts are accurate and understandable in the target task. Through the above settings, the user can intuitively see the result of the first text data after language conversion on the operation interface. The operation interface can display the second text data in natural language, allowing the user to intuitively understand the task being performed by the system and clearly see the converted second text data. This facilitates the determination of the correctness of the second text data and timely adjustment or correction of the second text data, which is convenient for subsequent generation of target prompts.

[0064] Step S206: In response to the first release command applied to the operation interface, display the first release result corresponding to the first release command on the operation interface.

[0065] The first published result is used to characterize the result of generating target prompt words based on the second text data. The target prompt words are used to guide the question-answering processing model to perform the target task.

[0066] The aforementioned first release instruction can refer to an instruction generated by the user interacting with the operation interface through clicking, selecting, or using language, which is used to control the release of target prompt words. The form of the first release instruction can be determined according to the specific interaction method between the user and the operation interface, and is not limited here.

[0067] The aforementioned target prompt words can refer to keywords or phrases used to guide the question-answering processing model to perform specific tasks. In this application, the target prompt words can be a single prompt word or a set of prompt words that can be dynamically adjusted according to user instructions and feedback from the question-answering processing model. Based on the target prompt words, the question-answering processing model can execute sequential logic, judgment logic, and loop logic, etc. The target prompt words can be determined according to actual needs, and are not limited here.

[0068] In one optional embodiment, target prompts corresponding to the target task can be generated based on the second text data. Users can publish the generated target prompts. After the user issues the first publishing command, the interface can display the first publishing result, i.e., the generated target prompts. The target prompts can help users better guide the question-answering processing model to perform the target task. Through the above settings, users can obtain target prompts through simple command operations without needing to understand the system's working principles. Displaying the first publishing result on the interface also helps users quickly view and share the generated prompts, thereby better utilizing these prompts for subsequent tasks. Users can directly generate target prompts based on their familiar natural language, enabling intuitive and user-friendly operation with the system. The user interface allows for interaction without requiring users to understand or master the complex programming languages ​​that the question-answering model can handle. Users can focus solely on the logic of the target task, reducing the learning and programming time costs associated with generating target prompts. These prompts help the question-answering model understand the user's intent, enabling it to answer questions or perform other tasks more effectively. Through these prompts, the model clearly understands the specific task the user wants to perform, thus better locating and handling the problem. Furthermore, the prompts provide additional contextual information, helping the model understand the background and context of the problem, improving the accuracy of problem comprehension. They also guide the model to perform more precise searches and inferences, enhancing the performance and efficiency of the question-answering model.

[0069] The solution provided by the above embodiments of this application, in response to a first input command applied to the operation interface, displays first text data in natural language on the operation interface, wherein the first text data is used to describe the execution process of the target task performed by the question-answering processing model; in response to a conversion command applied to the operation interface, displays a conversion window on the operation interface, wherein the conversion window displays second text data in the target language, the second text data being text data obtained by language conversion of the first text data, and the target language being used to represent the language corresponding to the domain to which the target task belongs; in response to a first release command applied to the operation interface, displays a first release result corresponding to the first release command on the operation interface, wherein the first release result is used to represent the result of generating target prompt words based on the second text data, and the target prompt words are used to guide the question-answering processing model to perform the target task. It is noteworthy that the system can automatically generate prompt words by responding to user input commands, conversion commands, and release commands. Users can input natural language to describe the target task of the question-answering processing model, and the system can display the task execution process based on the user's input text data to help users confirm task requirements. Users can use conversion commands to convert the displayed text data into text data in the target language to obtain the task requirements and determine the linguistic context of the required prompt words. The system can respond to the user's first release command to generate and display prompt word results based on the target language text data, guiding the question-answering processing model to execute the target task. Specifically, based on the conversion window, users can understand the tasks being performed by the system in real time and intuitively, clearly see the converted second text data, facilitate the determination of the correctness of the second text data, and make timely adjustments or corrections to the second text data. The generated prompt words can meet user needs and achieve accurate generation of prompt words. Users can directly describe the target task based on their familiar natural language and interact with the system's intuitive and user-friendly interface through simple command operations. Users do not need to understand or master the complex programming languages ​​that the question-answering processing model can handle. Users can only focus on the logic of the target task itself, reducing the learning cost and programming time cost of generating target prompt words, achieving efficient generation of prompt words, and thus solving the technical problems of low accuracy and efficiency in generating prompt words in related technologies.

[0070] In the above embodiments of this application, the method further includes: assembling the first text data into a first prompt word based on a preset prompt word template; and using the first prompt word to guide the text conversion model to generate the second text data.

[0071] The aforementioned preset prompt word templates can refer to some predefined templates or rules used to guide the processing and assembly of natural language text input by users for subsequent processing or generation. Preset prompt word templates can include keyword extraction templates, sentence transformation templates, question templates, etc., and can be designed and used according to specific needs and scenarios. By using preset prompt word templates, the entire text processing and generation process can be made more standardized and efficient.

[0072] The text conversion model described above can convert user-input natural language into a target language that the question-answering processing model can understand and process. It can also convert first text data into second text data. Specifically, the text conversion model can understand and analyze user-input natural language, extract key information and intent from the natural language, and convert the understood text into DSL or other programming languages ​​so that the question-answering processing model can understand and execute it. The specific type of text conversion model can be determined according to actual needs and is not limited here.

[0073] In one optional embodiment, a text generation model can be used to convert the first text language into second text data. The first text data can be assembled into a first prompt word based on a preset prompt word template. The preset prompt word template can be designed and constructed according to specific application scenarios. The preset prompt word template may include keywords or sentence structures. The user-input natural language data can be assembled according to the preset prompt word template to generate the first prompt word. Based on existing templates or rules, the user-input natural language text can be processed to extract key information, and this information can be assembled into the required prompt words or question-and-answer format text data according to the preset prompt word template. This process can include... This includes natural language processing techniques such as part-of-speech tagging, named entity recognition, and syntactic analysis. Specifically, a preset prompt word template can be set, which can include key information to be filled in, such as names, locations, and times. This information can then be assembled into the required prompt words or question-and-answer format text data based on the template. This process can include part-of-speech tagging, which determines the part of speech of each word in the sentence, such as nouns, verbs, and adjectives, helping to ensure that the generated prompt words conform to grammatical rules. Named entity recognition can also be performed to identify entity information in the text, such as names, place names, and organization names, ensuring that the generated prompt words contain accurate entity information. Furthermore, it can also... Sentence grammar analysis, or the analysis of sentence structure, can include the relationships between sentence components and the subject-verb-object structure. Through grammar analysis, it can be ensured that the generated prompt words conform to grammatical rules and are fluent and easy to understand. For example, to generate a prompt word about travel, the preset prompt word template could be "time + location + activity," such as "tomorrow + location A + visit attraction B." In the above process, part-of-speech tagging can be used to determine that "tomorrow" is a time word and "location A" is a location name. Then, named entity recognition is used to identify entity information, and finally, grammar analysis is used to ensure that the generated prompt words conform to grammatical rules. The final generated prompt word might be "Tomorrow I will go to location A to visit attraction B." Using the first prompt word to guide the text conversion model to generate the second text data: The first prompt word can be used as input, and the text generation model can generate the second text data that meets the preset target. The first prompt word, which has been preprocessed and assembled, can be used as input to guide the text conversion model, such as generative adversarial networks or recurrent neural networks, to generate the second text data in a specific language. Through the above steps, the first text language can be converted into the second text data. Using the preset prompt word template can help improve the text generation effect. Assembling the first text data through the preset prompt word template can effectively guide the model to focus on the important information in the text and avoid generating meaningless or redundant content.Using the generated first cue words to guide the text conversion model in generating second text data can make the generated text more closely match the expected theme or content, improve the text's coherence and consistency, and thus effectively improve the accuracy and readability of the generated text, making it more in line with user needs and expectations.

[0074] In the above embodiments of this application, the method further includes: obtaining a target prompt word template that matches the target prompt word, and an interface plugin that matches the target task, wherein the interface plugin is used to represent the interface called by the question-answering processing model during the execution of the target task; and assembling the second text data into a target prompt word based on the target prompt word template and the interface plugin.

[0075] The aforementioned target prompt template can refer to a template that contains keywords or phrases that match the target prompt. These keywords or phrases can help the question-answering model better understand the user's intent and generate appropriate answers or prompts. The target prompt template can include question patterns, keywords, grammatical structures, etc., which can be determined according to actual needs and are not limited here.

[0076] The aforementioned interface plugins can refer to the interfaces called by the question-answering processing model during the execution of the target task. They can help the model interact with external systems or resources. Interface plugins can be plugins used to obtain external data or plugins used to perform specific operations. They can be determined according to actual needs and are not limited here.

[0077] In one optional embodiment, a target prompt template matching the target prompt can be obtained. The target prompt template may include information such as the structure of the question, the location of keywords, and possible answer types. Next, an interface plugin matching the target task can be determined. The interface plugin can be used to represent the interface called by the question-answering processing model during the execution of the target task, such as a natural language processing model or knowledge graph. During the assembly of the second text data into target prompts, the second text data can be processed and adjusted according to the requirements of the target prompt template and the interface plugin to generate target prompts that meet the requirements. Specifically, a corresponding target prompt template can be determined based on the target prompts. The target prompt template may include placeholders, which can be used to fill in specific text content. Then, a suitable interface plugin can be selected according to the needs of the target task. The interface plugin is used to represent the interface called by the question-answering processing model during the execution of the target task. The interface plugin can be a pre-trained model, an application programming interface, etc. Finally, the second text data can be assembled into the target prompt word according to the target prompt word template and the interface plugin. This can include filling in the placeholders in the template and calling the interface plugin for text processing. Through the above steps, the process of obtaining the target prompt word template and interface plugin that match the target prompt word and assembling the second text data into the target prompt word can be realized. This can process the text data more efficiently, thereby meeting the requirements of the target task. Through the above steps, the second text data can be transformed into target prompt words, providing more accurate and effective information for the execution of the question-answering processing model. Through the above steps, target prompt words that match the target task can be effectively generated, thereby improving the performance and effectiveness of the question-answering processing model in a specific domain.

[0078] In the above embodiments of this application, after displaying the conversion window on the operation interface, the method further includes: responding to a confirmation command acting on the conversion window, displaying an original flowchart on the operation interface, wherein the original flowchart is a flowchart obtained by converting the second text data, wherein the original flowchart contains multiple process nodes and edges between different process nodes, wherein different process nodes are used to represent different execution steps of the target task, and edges are used to represent the execution order between different process nodes; responding to a first release command, displaying a first release result on the operation interface, wherein the first release result is used to represent the result of generating target prompt words based on the original flowchart.

[0079] The aforementioned original flowchart can refer to a visual flowchart obtained by directly converting the second text data. It is used to show the execution steps and execution order of the target task. The original flowchart can help users understand the execution process of the task more intuitively, improve users' understanding of task execution and visualization effect.

[0080] In one optional embodiment, when the system receives a user's confirmation instruction, it can display the original flowchart on the operation interface. The original flowchart is obtained by converting the second text data, such as the language specific to the domain of the question-and-answer processing model. The original flowchart can contain multiple process nodes and edges. Process nodes can represent different execution steps of the target task, and edges can represent the execution order between different process nodes. When the system receives a first release instruction, it can display the first release result on the operation interface. When the user needs to release a result or content, the system will respond to the user's release instruction and display the first release result on the operation interface to meet the user's needs. In the process of generating target prompt words, the system can generate appropriate prompt words to guide the user to complete the task or obtain information based on the user's instructions and needs, combined with the conversion and processing of text data. This can improve the user experience and make it more convenient and faster for the user to complete the target prompt word generation process.

[0081] This application converts text data into a graphical representation using original flowcharts, enabling users to more intuitively understand the task execution process and improving visualization. The original flowcharts can be displayed through an interface, allowing users to interact with them, such as clicking nodes to view detailed information and dragging nodes to rearrange them, thus enhancing user experience and ease of operation. Converting text data into original flowcharts helps users understand the task execution process more quickly, improving work efficiency and accuracy. It enhances users' understanding and visualization of the task execution process and facilitates users' intuitive and visual adjustments to the content of the second text data.

[0082] In the above embodiments of this application, the method further includes: responding to a modification instruction applied to the operation interface, displaying a target flowchart on the operation interface, wherein the target flowchart is a flowchart obtained by modifying the original flowchart based on the modification instruction; responding to a second release instruction applied to the operation interface, displaying a second release result corresponding to the second release instruction on the operation interface, wherein the second release result is used to characterize the result of generating target prompt words based on third text data, and the third text data is text data obtained by converting the target flowchart into a target language.

[0083] The aforementioned modification instructions can refer to instructions generated by the user interacting with the operation interface through clicking, selecting, or using language, which are used to modify the original flowchart. The form of the modification instructions can be determined according to the specific interaction method between the user and the operation interface, and is not limited here.

[0084] The aforementioned third text data can refer to text data corresponding to the content of the target flowchart. Users of this application only need to interact with the operation interface to directly adjust or modify the original flowchart. The adjusted or modified flowchart serves as the target flowchart. The system can automatically convert the target flowchart into third text data in the backend. The third text data can be determined according to actual needs and is not limited here.

[0085] In one optional embodiment, when a user modifies or edits the target flowchart, the system can respond promptly and display the modified target flowchart on the operation interface. The user can intuitively modify the flowchart without paying too much attention to the target language itself and can intuitively see the changes made. When the user needs to further publish and generate a second publication result, the system can display the corresponding result on the operation interface according to the second publication instruction. The second publication result is obtained by processing and analyzing the third text data and is used to generate the final target prompt words. Through the above steps, the system can more intelligently generate the required third text data according to the user's operation and needs. The user can adjust the target flowchart by modifying the instructions, which can meet the user's specific needs and goals. The above interactive generation method can improve the user experience and also increase the system's flexibility and adaptability.

[0086] This application allows users to directly adjust or modify the original flowchart simply by interacting with the user interface. The adjusted or modified flowchart serves as the target flowchart, which the system can automatically convert into third-party text data in the backend to generate target prompts. Users do not need to focus on the complex target language itself to adjust or modify the target task flow. Users do not need to understand or master the complex programming languages ​​that the question-answering model can handle. Users only need to focus on the logic of the target task itself and directly adjust or modify the original flowchart. This reduces the learning cost and programming time cost of generating target prompts for users, and allows for timely and convenient indirect adjustment or modification of the logic of the target task described in the target language, thus improving the efficiency of target prompt generation.

[0087] In the above embodiments of this application, before responding to the first release command applied to the operation interface, the method further includes: responding to the debugging command applied to the operation interface and displaying a debugging window on the operation interface; responding to the second input command applied to the debugging window and displaying query text data corresponding to the second input command in the debugging window; and displaying response text data corresponding to the query text data in the debugging window, wherein the response text data is text data generated by using the second text data to guide the question-and-answer processing model to process the query text data.

[0088] The aforementioned debugging instructions can refer to instructions generated by the user interacting with the operation interface through clicking, selecting, or using language, which are used to debug the second text data. The form of the debugging instructions can be determined according to the specific interaction method between the user and the operation interface, and is not limited here.

[0089] The aforementioned debugging window can be a window that displays the debugging process to the user. The debugging window can display the query text data corresponding to the second input command, the corresponding flowchart during the debugging process, or the response text data corresponding to the query text data. The form of the debugging window can be determined according to actual needs and is not limited here.

[0090] The aforementioned second input instruction can be a user-input instruction or information used to guide the question-answering processing model in processing the query text data. The second input instruction can include specific requirements for the topic, details of the question, or content that needs special processing. By responding to the second input instruction, the system displays the corresponding query text data and response text data in the debugging window, which can help users understand the model's processing process and results more clearly. It also provides users with a more intuitive operating interface. By guiding the question-answering processing model to process the query text data and generate response text data, the generated target prompts can be more accurate and targeted.

[0091] In one optional embodiment, in response to a debugging command applied to the user interface, a debugging window can be displayed on the user interface. The debugging window can be used to input commands, view data, and perform other operations. When a second input command is received applied to the debugging window, the query text data corresponding to the second input command can be displayed in the debugging window. The query text data can be a question or request raised by the user. The corresponding response text data can be displayed in the debugging window. The response text data can be the text data generated by guiding the question-and-answer processing model to process the query text data using the second text data. The text data can be an answer generated based on the user's question, or prompt words, etc. Through the above steps, convenient and flexible debugging of the second text data can be achieved.

[0092] In the above embodiments of this application, the method further includes: parsing the second text data to determine multiple process nodes contained in the second text data; determining prompt word templates corresponding to the multiple process nodes based on the node types of the multiple process nodes; assembling the second text data into prompt words corresponding to the multiple process nodes based on the prompt word templates corresponding to the multiple process nodes; and sequentially using the prompt words corresponding to the multiple process nodes to guide the question-answering processing model to process the query text data to obtain the response text data.

[0093] In one optional embodiment, the second text data can be parsed to determine multiple process nodes contained within it. Each process node refers to a step or stage that needs to be completed in a task or objective, and each node may have a specific function and role. Based on the node types of the multiple process nodes, prompt word templates corresponding to these process nodes can be determined. These prompt word templates can be templates used when generating target prompt words and may include fixed keywords and variable parts. The prompt word templates can help generate more accurate and effective target prompt words. Next, based on the prompt word templates corresponding to the multiple process nodes, the second text data can be assembled into prompt words corresponding to the multiple process nodes. This can be achieved by matching and replacing key information in the text data with the prompt word templates, thereby generating target prompt words that meet the requirements. Finally, the prompt words corresponding to the multiple process nodes can be used sequentially to guide the question-and-answer processing model to process the query text data and obtain response text data. Based on the generated target prompt words, the question-and-answer processing model can better understand the user's query and provide a corresponding response. Through the above process, the generation and application of target prompt words can be realized, improving the accuracy and efficiency of the question-and-answer processing model.

[0094] Specifically, a process DSL can be parsed to obtain different types of nodes, such as start, information retrieval, function call, branch judgment, generation, or end nodes. Different node types correspond to different operations and logic. Different types of prompts can be extracted and assembled (Prompt Assemblyprompt, or simply prompt). Based on the parsed multiple process nodes, prompt word templates corresponding to different process nodes can be determined. Specifically, the node types of different process nodes can be determined, such as start nodes, decision nodes, and end nodes. Then, for different types of nodes, corresponding prompt word templates can be determined. For start nodes, prompt word templates such as "Please begin your operation" or "Please follow the prompts to proceed to the next step" can be determined. For decision nodes, different prompt word templates can be designed based on different choices. For example, if it is a yes / no selection node, the prompt word can be designed as "Please make a selection: yes / no"; if it is a selection node with multiple options, the prompt word can be designed as "Please select your operation: option 1 / option 2 / option 3"; for end nodes, some end prompt words can be designed, such as "Operation completed, thank you for your cooperation" or "Task completed, please close the page." Through the above steps, prompts can be designed based on different process nodes. The system determines prompt word templates corresponding to multiple process nodes. In practical applications, these templates can be flexibly adjusted according to specific application needs and process design to improve user experience and operational smoothness. The prompt word templates can be customized and adjusted based on actual conditions, without limitation here. Based on these prompt word templates, the second text data can be assembled into prompt words corresponding to multiple process nodes. This generates corresponding prompt words for each node to guide the user through the preset process. Finally, the prompt words corresponding to multiple process nodes can be used to guide the question-and-answer processing model to process the query text data and obtain the response text data. Through these steps, it can be ensured that the question-and-answer processing model can operate according to the preset process and generate expected response text data. This application, by parsing the second text data, determining the node type, generating prompt word templates, assembling prompt words, and guiding the question-and-answer processing model, can achieve the generation of target prompt words and process guidance, improving user experience and system efficiency.

[0095] In the above embodiments of this application, the question-and-answer processing model is guided to process the query text data and obtain the response text data by sequentially using prompt words corresponding to multiple process nodes. This includes: determining the target process node to be executed among multiple process nodes; using prompt words corresponding to the target process node to guide the question-and-answer processing model to process the historical processing results and obtain the target processing result corresponding to the target process node, wherein, when the target process node is the first executed process node among multiple process nodes, the historical processing result is the query text data; when the target process node is not the first executed process node, the historical processing result is the processing result corresponding to the process node executed before the target process node; when the target process node is not the last executed process node among multiple process nodes, the process node executed after the target process node is taken as the target process node, and the step of guiding the question-and-answer processing model to process the historical processing results and obtain the target processing result by using prompt words corresponding to the target process node is repeated until the target process node is the last executed process node; when the target process node is the last executed process node among multiple process nodes, the target processing result is determined to be the response text data.

[0096] In one optional embodiment, the target process node to be executed among multiple process nodes can be determined. By identifying the prompt words corresponding to each process node, the question-answering processing model can be guided to process the query text data to determine the current process node. The prompt words corresponding to the target process node can be used to guide the question-answering processing model to process the historical processing results to obtain the target processing result corresponding to the target process node. If the target process node is the first process node to be executed among multiple process nodes, the historical processing result can be the query text data. If the target process node is not the first process node to be executed, the historical processing result can be the processing result corresponding to the process node executed before the target process node. The prompt words corresponding to the target process node can be repeatedly used to guide the process. The question-answering processing model processes historical results until the target process node is the last executed process node. If the target process node is the last executed process node among multiple process nodes, then the target processing result is determined to be the response text data. This application can produce the output results required for each node type and the next node sequence number, and store these results in the context parameter list for use by subsequent nodes. Through the cyclic execution of the step iterator, the entire process can continue until the end of execution. This process design can ensure that the question-answering processing model generates target prompt words step by step according to the set process node order, improving generation efficiency and accuracy. Through the above steps, the question-answering processing model can be effectively guided to process the query text data and finally obtain the response text data.

[0097] In the above embodiments of this application, in response to a viewing command applied to the debugging window, the method further includes: displaying a process execution window on the operation interface, wherein the process execution window displays the original flowchart and the execution status of different process nodes in the original process, wherein the execution status is used to characterize whether the corresponding process node has been executed successfully.

[0098] In one optional embodiment, a process execution window can be displayed on the user interface. The process execution window can display the original flowchart and the execution status of different process nodes in the original process. The design of the process execution window can help users clearly understand the current execution status of the process and the execution status of each node. The execution status can also be represented by different colors or icons, such as green for success, red for failure, and yellow for progress. Users can clearly and intuitively see which nodes have been executed successfully, which nodes are still in progress, or which have failed. The design of the process execution window allows users to more intuitively understand the execution status of the entire process, helping users to better grasp the progress of the process and the solution of problems, and improve the efficiency of generating target prompts.

[0099] In the above embodiments of this application, in response to a first input instruction applied to the operation interface, the method further includes at least one of the following: in response to a template selection instruction applied to the operation interface, displaying a set of prompt word templates corresponding to the template selection instruction on the operation interface, wherein different prompt word templates in the template set are used to assemble different text data into prompt words; in response to a model selection instruction applied to the operation interface, displaying the type corresponding to the question-answering processing model on the operation interface; in response to a plugin selection instruction applied to the operation interface, displaying an interface plugin that matches the target task on the operation interface.

[0100] In one optional embodiment, in response to a template selection instruction applied to the user interface, the system can select a corresponding template according to the user's instruction, and then display a set of prompt word templates corresponding to the template selection instruction on the user interface. The prompt word template set may include different prompt word templates, and each prompt word template can be used to assemble different text data into prompt words. By selecting different templates, the system can generate prompt words with different styles and content according to the user's needs. In response to a model selection instruction applied to the user interface, the system can respond to the user's model selection instruction and display the type of question-answering processing model on the user interface. The model may be built based on machine learning, deep learning, and other technologies to process questions and generate answers. According to the model type selected by the user, the system can call different algorithms and models to generate more accurate prompt words that meet the user's needs. In response to a plugin selection instruction applied to the user interface, the system can display interface plugins that match the target task on the user interface according to the user's plugin selection instruction. Interface plugins can provide additional functions and services to help users better generate and customize prompt words. By selecting appropriate plugins, users can extend the functionality of the system to achieve more complex and diverse prompt word generation needs.

[0101] This application proposes a method for constructing service domain process automation intelligent agents based on a large model. This method achieves low-cost configuration, requiring only natural language description of the process without code development or interface calls. It is simple to operate and achieves full automation. It can generate DSL and flowcharts with one click from the large model, enabling fully automatic operation. It has controllable construction, allowing controllable processes to be built based on the DSL. While flexibly generating from the large model, it ensures the accuracy of process execution and can be provided to front-line customer service personnel.

[0102] Agent developers can simply write the process in natural language and input it into the large model. The large model will then convert this natural language version of the process into a domain-specific language (DSL). In practice, this solution can use a custom-structured JSON as the DSL version, or it can use other domain-specific languages ​​such as YAML, XML, or a completely custom syntax. The DSL is then parsed and rendered into a flowchart. After the agent developers confirm and fine-tune the flowchart, the final version will serve as the basis for agent process execution. Subsequently, the adjusted flowchart will be converted back into a DSL and then executed within the process execution framework. The process execution framework will assemble plugins (Schema, Prompt, etc.) and then provide the complete Prompt to the large model to answer questions raised by frontline customer service, ultimately generating corresponding solutions. The plugins can be developed by plugin developers, while the questions can be input by frontline customer service personnel.

[0103] Figure 3 This is a schematic diagram of an optional target prompt word generation process according to an embodiment of this application, such as... Figure 3 As shown, Figure 3 As shown, agent developers can use natural language to describe the process, obtaining first text data. This first text data is then input into a large model (LLM), which can convert the first text data into second text data. Optionally, the large model (LLM) can convert the second text data into an original flowchart. Agent developers can adjust the original flowchart to obtain the final adjusted and confirmed target flowchart. The large model (LLM) can then convert the target flowchart into third text data. A process execution framework can be generated based on the second / third text data. Plugin developers can provide the organization and relationship (schema) of plugins and data, assembling the process execution framework and the organization and relationship of plugins and data. Combined with questions raised by frontline customer service personnel, the final target prompt words can be generated.

[0104] Figure 4 This is a schematic diagram of another optional target prompt word generation process according to an embodiment of this application, such as... Figure 4As shown, the specific implementation process of the technical solution proposed in this application may include a process creation and target language conversion process, and a process and target language execution process. The process creation and target language conversion process may include the agent developer inputting a target task described in natural language to obtain first text data, and then converting the first text data into second text data described in the target language based on the large model. The process and target language execution process may include sequentially traversing and executing the start node, information acquisition node, function call node, branch judgment node, generation node, and end node based on a step iterator. The start node can jump to the next node. The information acquisition node can acquire information fragments provided by the user, such as questions provided by frontline customer service personnel; acquire global parameters from the system's global parameter list; provide the acquired global parameters to the large model (LLM); and acquire new global parameters from the user, update the global parameter list, and jump to the next node. Here, global parameters may refer to the set of prompt word templates corresponding to the template selection instruction, the interface plugins matching the type of the question-answering processing model and the target task, etc. Function call nodes can assemble user-provided information fragments based on documents or files defining function interface structures and specifications; retrieve global parameters from the system's global parameter list and provide them to the Large Model (LLM); call the corresponding function interface based on the parameter information required by the function interface; after the function interface completes the operation requested by the client, it can return the execution result to the client and jump to the next node. The parameter information required by the function interface can be JSON data, etc., without limitation. Branch decision nodes can process the branch structure in text data, perform branch decisions, retrieve global parameters from the system's global parameter list, provide them to the Large Model (LLM), and jump to the next node. Generation nodes can assemble the process execution framework, plugins, and data organization methods and relationships, and generate the final target prompt words based on questions raised by frontline customer service personnel. They can also retrieve global parameters from the system's global parameter list, provide them to the Large Model (LLM), and jump to the next node. End nodes can execute the process stop step.

[0105] Optionally, the process creation and target language conversion can include the following: when the agent developer inputs a natural language description of the process, it can be assembled into a new process prompt based on a process conversion template; then, the process prompt can be converted through a large model with a large number of parameters to generate a process domain-specific language (DSL). Afterward, the process DSL can be rendered as a flowchart; and after the agent developer fine-tunes and edits the flowchart, it can finally be converted back into the final process DSL.

[0106] The process of executing the process and the target language may include passing the final process DSL to the process execution framework for execution. Within the process execution framework, step iterators can be set up. These iterators can parse the process DSL to obtain different types of nodes, such as start, information retrieval, function call, branch judgment, generation, or end nodes. For different node types, different types of prompts can be extracted and assembled. Then, the actual execution process is carried out through a large model. Here, a large model with lightweight parameters can be used. The specific descriptions of various node types are as follows:

[0107] The start node can represent the beginning of the process. It can be a virtual node or have no real information, but it is required in the DSL. The "id" field starts from 0 and has "name" and "description" fields. "name" is "start" and "description" is "start node". The next node "next_step" points to the step with id 1.

[0108] The information acquisition node can be used to obtain information from customers, such as instance ID, account ID, etc. It can be done according to the specific process description and the required information is converted into the "parameters" field.

[0109] The function call node can be the API that needs to be called from a certain plugin. It depends on the plugin or API name described in the specific process, and the required plugin name is placed in the "api_name" field.

[0110] Branch decision nodes can be nodes that require branching decisions. If the branch decision needs to refer to the result of a function call, the "api_result" field needs to be added, and the plugin's API name needs to be filled in. If the decision is based only on a certain condition or text description, the "api_result" field is not needed. The result of the branch decision needs to be specified through the "switch" field, which is an array containing different "case" situations. In different situations, different "next_step" needs to be executed, and the sequence number of the next step needs to be written.

[0111] The generated node can be used to generate a script to reply to customers. In this case, if it depends on the execution result of a certain plugin API, then "api_result" is needed. Otherwise, this field is not needed. If it needs to refer to a certain script, then the "reference" field is needed to display the required script, document, and other reference information.

[0112] The end node can be a virtual node without any substantial information, but it must exist. "id" is the sequence number of the last step, "name" is "end", and "description" is "end node". The "next_step" field can be omitted to serve as the last node.

[0113] Finally, the large model can produce the required output results for each node type and the next node number, and then give them to the step iterator. The step iterator will store these results in the corresponding context parameter list or give them to subsequent downstream nodes for use, and continue to execute the next node. This process repeats until the entire process is completed.

[0114] Compared to traditional intelligent agent configuration platforms, the technical solution proposed in this application eliminates the need for traditional process orchestration components when constructing intelligent agent processes. Users are no longer required to build complete process links or fill in various parameters and API details; they only need to describe the process text using "natural language." It enables one-click flowchart conversion, using a large model with a large number of parameters to read user-written natural language descriptions and convert them into a specific DSL language. The DSL is then rendered into a flowchart. If the flowchart does not meet requirements, the natural language description can be rewritten for conversion, or the flowchart can be fine-tuned. Furthermore, it enables controllable flowchart execution. After the flowchart is finalized, a lightweight large model can execute the steps sequentially according to the agreed-upon steps, ensuring controllable completion of the entire process. The execution process also uses the large model, allowing for control between nodes but greater flexibility within nodes, eliminating the need for the complete parameter configuration and link assembly required in traditional process orchestration.

[0115] Optionally, users can create and execute DSL-based process agents through actual operation. Users can start creating an agent by clicking the "Create" button on the interface, filling in basic information, and entering "process text" in the "Process Flow" input box (natural language is acceptable). Users can choose from built-in Prompt templates, model types, and corresponding plugins. Clicking the "Convert" button on the interface activates the automatic conversion mode, converting the text-described process into the target language. The large model can generate process descriptions in domain-specific languages ​​and also generate flowcharts rendered using DSL. Clicking the "Debug" button allows users to debug the process. The agent is started through a dialogue, during which the process steps are displayed in real-time, and the output results are as expected. The process automatically exits upon completion. Clicking the "Flowchart" icon in the upper right corner of the dialogue displays the execution status of the tree-structured flowchart. Finally, clicking the "Publish" button on the interface officially publishes the agent for external use.

[0116] Figure 5 This is a schematic diagram of an optional user interface according to an embodiment of this application, such as... Figure 5 As shown, the operation interface 502 may include a first text data display area 506, a flowchart display area 508, a system configuration area 504, and a function selection area 510. The first text data display area can display the first text data in natural language input by the user and allows interaction with the user, enabling real-time modification of the first text data. The flowchart display area can display the original / target flowchart automatically converted by the system and allows interaction with the user, enabling real-time modification of the flowchart. The system configuration area can be equipped with multiple display boxes, allowing the user to generate template selection instructions, model selection instructions, and plugin selection instructions based on filling in or selecting, and selecting the target prompt words. The system configuration area can display the set of prompt word templates corresponding to the template selection instruction, the type of question-and-answer processing model, and the interface plugins that match the target task. The content displayed in the system configuration area can also be set according to actual needs, and is not limited here. The function selection area can be set with multiple selection buttons or options, and can be set with functions such as save, publish, debug, and convert. Users can save the current version, publish the current content, debug the current content, and convert between text data and flowcharts. The settings of each area in the above operation interface can be determined according to actual needs, and are not limited here.

[0117] Figure 6This is a schematic diagram of an optional debugging window according to an embodiment of this application, such as... Figure 6 As shown, the debugging window 602 can display a left-side display window 604 and a first right-side display window 606 with left-right contrast. The left-side display window can display the text data entered by the user and the system configuration information, while the first right-side display window can display the generated target prompt page. The user can adjust the content and components in the target prompt display page. The display content of the left-side display window and the first right-side display window can change in real time according to the user's operation, which enhances the interactivity with the user and improves the user's debugging efficiency for the target prompt. The settings of each area in the above debugging window can be determined according to actual needs and are not limited here.

[0118] Figure 7 This is a schematic diagram illustrating an optional response to a viewing command applied to a debug window according to an embodiment of this application, such as... Figure 7 As shown, in response to a viewing command applied to the debug window, the debug window 602 can display a left-side display window 604 and a second right-side display window 702 with left-right contrast. The left-side display window can display the text data entered by the user and the system configuration information, while the second right-side display window can display the flowchart corresponding to the text data and the execution status of different process nodes in the process. The second right-side display window is displayed in response to a viewing command applied to the debug window. This design helps users clearly understand the current execution status of the process and the execution status of each node, allowing users to more intuitively understand the execution status of the entire process, helping users better grasp the progress of the process and the solution of problems, and improving the efficiency of generating target prompts.

[0119] The technical solution proposed in this application enables low-cost process configuration. Compared to traditional process processing methods, it eliminates the need for extensive development resources or complex maintenance costs of configuration platforms. It requires only pure natural language descriptions, eliminating the need for any code development or process orchestration capabilities, making process configuration simple and easy to use. This reduces the technical requirements for users, who no longer need in-depth understanding of prompts, parameter configurations, API calls, and link connections. It achieves fully automated, code-free operation. Users can generate domain-specific language (DSL) and flowcharts with a single click based on pre-written natural language descriptions using a large model. This fully automated operation allows users to focus solely on the process itself, without needing to spend excessive effort on aspects outside the process. Furthermore, this technical solution implements DSL-based process construction and proposes a controllable, personalized intelligent agent process construction method based on a domain-specific language. This method ensures process controllability while flexibly generating data using a large model, resulting in low configuration costs while guaranteeing the stability and accuracy of process operation.

[0120] According to an embodiment of this application, a text processing method is also provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0121] Figure 8 This is a flowchart of a text processing method according to an embodiment of this application, such as... Figure 8 As shown, the specific steps may include the following:

[0122] Step S802: In response to the input command applied to the interactive interface, display the current query text on the interactive interface.

[0123] In one alternative embodiment, the system can respond to instructions input by the user on the interactive interface, which may include recognizing the user's intent and needs, and then displaying the current query text on the interactive interface so that the user can clearly see the question or information being asked. Specifically,

[0124] In responding to input commands on the interactive interface, the system can recognize and parse the user's input text to determine the user's intent and needs. This involves natural language processing technologies, including text analysis, semantic understanding, and entity recognition. The system can analyze the user's input text, extract key information, and then generate corresponding prompts or answers based on this information. By displaying the current query text on the interactive interface, the system can show the generated prompts or answers, allowing the user to intuitively see the system's feedback. This helps the user understand the system's response and further interact with the system. Displaying the current query text helps the user confirm whether their question or need has been correctly understood and whether the system's answer meets their expectations. By displaying the current query text, the system can help the user better understand the system's response, promoting effective communication and exchange.

[0125] Step S804: In response to the processing instruction applied to the interactive interface, display the current response text corresponding to the current query text on the interactive interface.

[0126] The current response text is generated by the question-answering model using target prompts, which are prompts generated using the method described above.

[0127] In one optional embodiment, after the processing instruction is transmitted to the interactive interface, the system can select an appropriate response text based on the target prompt words. Specifically, the system can receive the query text input by the user and process it according to the preset processing instructions. The system can guide the question-and-answer processing model to process the current query text based on the target prompt words and generate the corresponding response text. The response text can be displayed on the interactive interface for the user to view and understand. By using the target prompt words and the method for generating the target prompt words, the accuracy and efficiency of the question-and-answer processing model can be improved, the user interaction experience and system performance can be improved, and the system can better understand the user's intent and needs, thereby providing more accurate and personalized responses.

[0128] According to an embodiment of this application, a method for generating prompt words is also provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0129] Figure 9 This is a flowchart of an optional prompt word generation method according to an embodiment of this application, such as... Figure 9 As shown, the specific steps may include the following:

[0130] Step S902: Obtain the first text data of natural language.

[0131] The first text data describes the execution process of the target task performed by the question-answering processing model.

[0132] In one optional embodiment, text data describing the execution process of the question-answering processing model in performing the target task can be acquired so that the model can learn and generate appropriate prompts. The text data may include user-inputted questions or needs, system-returned answers or prompts, and dialogue data generated during user interaction with the system. By acquiring the first text data, the model can learn user habits, preferences, and needs, thereby better understanding user intentions and behaviors, and generating target prompts that better meet user expectations. This helps the model better understand user needs and behaviors, thereby generating prompts that are closer to user expectations, improving user experience and system performance.

[0133] Step S904: Perform language conversion on the first text data to obtain the second text data in the target language.

[0134] The target language is used to represent the language corresponding to the domain to which the target task belongs.

[0135] In one optional embodiment, a text conversion model can be used to perform language conversion, converting the first text data into the language corresponding to the domain of the target task. A neural machine translation model can be used to perform language conversion on the first text data to obtain the second text data. Alternatively, rule-based machine translation, statistical machine translation, or other methods can be used to perform language conversion on the first text data to obtain the second text data. The specific implementation method can be determined according to actual needs and is not limited here. Converting the first text data into the second text data can ensure that the generated target prompt words conform to the characteristics of the domain of the target task in terms of expression and word choice.

[0136] Step S906: Generate target prompt words based on the second text data.

[0137] Among them, the target prompt words are used to guide the question-answering processing model to perform the target task.

[0138] In one optional embodiment, target prompts can be generated based on the second text data after language conversion. These target prompts can guide the question-answering model to better perform the target task. The target prompts can be keywords, phrases, or sentences, used to provide key information or guidance about the target task, helping the question-answering model to more accurately understand and answer relevant questions. Through the above steps, target prompts for a specific target task can be obtained. The target prompts are closer to the domain of the target task in terms of language expression and information guidance, which helps to improve the performance and effectiveness of the question-answering model on the target task.

[0139] According to an embodiment of this application, a method for generating prompt words is also provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0140] Figure 10 This is a flowchart of another optional prompt word generation method according to an embodiment of this application, such as... Figure 10 As shown, the specific steps may include the following:

[0141] Step S1002: Obtain the first text data of natural language by calling the first interface.

[0142] The first interface includes a first parameter, the value of which includes first text data, which describes the execution process of the target task performed by the question-answering processing model.

[0143] In one optional embodiment, first text data in natural language can be obtained by calling a first interface. That is, text data describing the execution process of the question-answering processing model in performing the target task can be obtained through a pre-designed interface. The first interface may include one or more parameters, where the parameter value of the first parameter may be the first text data, i.e., text used to describe the execution process of the target task. In practical applications, the first text data may be natural language text containing a detailed description, or it may be a structured data format, such as JSON or XML. The text data may include background information of the task, execution steps, key instructions, etc., to help users or machines understand and execute the target task, and to provide basic information for the subsequent generation of target prompt words.

[0144] Step S1004: Perform language conversion on the first text data to obtain the second text data in the target language.

[0145] The target language is used to represent the language corresponding to the domain to which the target task belongs.

[0146] In one optional embodiment, a pre-trained text conversion model can be used to perform language conversion on the first text data to obtain second text data in the target language. Alternatively, by converting the input text into the target language, a neural machine translation model can be used to perform language conversion on the first text data to obtain the second text data. Rule-based machine translation, statistical machine translation, and other methods can also be used to perform language conversion on the first text data to obtain the second text data. Furthermore, language conversion can be performed on the first text data through manual translation, using specific language conversion algorithms, or by using a pre-defined mapping relationship between natural language and the target language. The specific implementation method can be determined according to actual needs and is not limited here. This ensures that the generated prompts are more aligned with the needs of the target domain, thereby improving the accuracy and effectiveness of the generated results. The target language can well represent the language characteristics and norms related to the domain to which the target task belongs, thus ensuring that the generated prompts meet the domain requirements. When performing language conversion, a suitable target language can be selected to ensure that the generated target prompts meet the domain requirements and produce good results.

[0147] Step S1006: Generate target prompt words based on the second text data.

[0148] Among them, the target prompt words are used to guide the question-answering processing model to perform the target task.

[0149] In one optional embodiment, target prompts are generated for the question-answering system to help the model better understand the user's intent and provide accurate answers. During the generation of target prompts, they can be generated based on second text data. This can be achieved by first acquiring a corpus containing a large amount of text data, and then using text processing techniques, such as word vector representation and text summarization, to extract key information and features. Machine learning or deep learning models can be used to train the algorithm for generating target prompts, enabling the generation of corresponding target prompts based on the input second text data. Generating target prompts allows the question-answering model to more accurately understand the user's question and intent, and provide more relevant and targeted answers. By guiding the model to use target prompts to perform specific tasks, the model's accuracy and efficiency can be improved, thereby enhancing the user experience and performance of the question-answering system.

[0150] Step S1008: Output the target prompt word by calling the second interface.

[0151] The second interface includes a second parameter, the value of which includes the target prompt word.

[0152] In one optional embodiment, machine learning or natural language processing techniques can be used to generate prompt words that meet specific needs. The generation process may include multiple steps. Specifically, the target prompt word can be output by calling a second interface, which can be a pre-trained model or algorithm used to generate the target prompt word. The second parameter can be used to specify the conditions or requirements for generating the target prompt word, such as length, theme, style, etc. By calling the second interface and passing in the parameter value of the second parameter, the system can obtain the target prompt word that meets the conditions. After obtaining the target prompt word, the system can further process or combine it as needed to finally generate prompt words that meet the requirements. This may involve various technologies such as text processing, semantic analysis, and generative models to ensure the quality and compliance of the generated prompt words. Through the above process, users can generate target prompt words more quickly and accurately.

[0153] According to an embodiment of this application, a prompt word generation apparatus for implementing the above-described prompt word generation method is also provided. Figure 11 This is a schematic diagram of a prompt word generation device according to an embodiment of this application, such as... Figure 11 As shown, the device includes: a first display module 1102, a second display module 1104, and a third display module 1106.

[0154] The system comprises the following components: a first display module 1102, which responds to a first input command applied to the operation interface and displays first text data in natural language on the operation interface, wherein the first text data describes the execution process of the target task performed by the question-answering processing model; a second display module 1104, which responds to a conversion command applied to the operation interface and displays a conversion window on the operation interface, wherein the conversion window displays second text data in the target language, which is text data obtained by language conversion of the first text data, and the target language is used to represent the language corresponding to the domain to which the target task belongs; and a third display module 1106, which responds to a first release command applied to the operation interface and displays a first release result corresponding to the first release command on the operation interface, wherein the first release result represents the result of generating target prompt words based on the second text data, and the target prompt words are used to guide the question-answering processing model to perform the target task.

[0155] The aforementioned device further includes a first assembly module and a generation module. The first assembly module is used to assemble the first text data into a first prompt word based on a preset prompt word template. The generation module is used to guide the text conversion model to generate second text data using the first prompt word.

[0156] The aforementioned device further includes an acquisition module and a second assembly module. The acquisition module is used to acquire a target prompt word template that matches the target prompt word and an interface plugin that matches the target task. The interface plugin is used to represent the interface called by the question-answering processing model during the execution of the target task. The second assembly module is used to assemble the second text data into a target prompt word based on the target prompt word template and the interface plugin.

[0157] The second display module is further configured to respond to a confirmation command applied to the conversion window and display the original flowchart on the operation interface. The original flowchart is a flowchart obtained by converting the second text data. The original flowchart contains multiple process nodes and edges between different process nodes. Different process nodes are used to represent different execution steps of the target task, and edges are used to represent the execution order between different process nodes. In response to the first release command, the first release result is displayed on the operation interface. The first release result is used to represent the result of generating target prompt words based on the original flowchart.

[0158] The device further includes a fourth display module, which responds to a modification command applied to the operation interface and displays a target flowchart on the operation interface, wherein the target flowchart is a flowchart obtained by modifying the original flowchart based on the modification command; and responds to a second release command applied to the operation interface and displays a second release result corresponding to the second release command on the operation interface, wherein the second release result is used to characterize the result of generating target prompt words based on third text data, and the third text data is text data obtained by converting the target flowchart into a target language.

[0159] The third display module is also used to respond to debugging commands applied to the operation interface and display a debugging window on the operation interface; respond to a second input command applied to the debugging window and display the query text data corresponding to the second input command in the debugging window; and display the response text data corresponding to the query text data in the debugging window. The response text data is the text data generated by using the second text data to guide the question-and-answer processing model to process the query text data.

[0160] The aforementioned device further includes a parsing module, used to parse the second text data and determine the multiple process nodes contained in the second text data; based on the node types of the multiple process nodes, determine the prompt word templates corresponding to the multiple process nodes; based on the prompt word templates corresponding to the multiple process nodes, assemble the second text data into prompt words corresponding to the multiple process nodes; and sequentially use the prompt words corresponding to the multiple process nodes to guide the question-answering processing model to process the query text data and obtain the response text data.

[0161] The parsing module is further used to determine the target process node to be executed among multiple process nodes; using the prompt words corresponding to the target process node, it guides the question-answering processing model to process the historical processing results to obtain the target processing result corresponding to the target process node. Specifically, if the target process node is the first executed process node among multiple process nodes, the historical processing result is the query text data; if the target process node is not the first executed process node, the historical processing result is the processing result corresponding to the process node preceding the target process node; if the target process node is not the last executed process node among multiple process nodes, the process node following the target process node is taken as the target process node, and the step of using the prompt words corresponding to the target process node to guide the question-answering processing model to process the historical processing results to obtain the target processing result is repeated until the target process node is the last executed process node; if the target process node is the last executed process node among multiple process nodes, the target processing result is determined to be the response text data.

[0162] The device also includes a fifth display module for displaying a process execution window on the operation interface. The process execution window displays the original flowchart and the execution status of different process nodes in the original process. The execution status is used to indicate whether the corresponding process node has been executed successfully.

[0163] The first display module is further configured to respond to a first input command applied to the operation interface. The method further includes at least one of the following: responding to a template selection command applied to the operation interface, displaying a set of prompt word templates corresponding to the template selection command on the operation interface, wherein different prompt word templates in the template set are used to assemble different text data into prompt words; responding to a model selection command applied to the operation interface, displaying the type corresponding to the question-answering processing model on the operation interface; and responding to a plugin selection command applied to the operation interface, displaying an interface plugin that matches the target task on the operation interface.

[0164] According to embodiments of this application, a text processing apparatus for implementing the above-described text processing method is also provided. Figure 12 This is a schematic diagram of a text processing apparatus according to an embodiment of this application, such as... Figure 12 As shown, the device includes: a first display module 1202 and a second display module 1204.

[0165] The first display module 1202 is used to respond to input commands applied to the interactive interface and display the current query text on the interactive interface; the second display module 1204 is used to respond to processing commands applied to the interactive interface and display the current response text corresponding to the current query text on the interactive interface, wherein the current response text is generated by the question-and-answer processing model using target prompt words to process the current query text.

[0166] According to an embodiment of this application, a prompt word generation apparatus for implementing the above-described prompt word generation method is also provided. Figure 13 This is a schematic diagram of an optional prompt word generation device according to an embodiment of this application, such as... Figure 13 As shown, the device includes: an acquisition module 1302, a conversion module 1304, and a generation module 1306.

[0167] The acquisition module 1302 is used to acquire first text data in natural language, wherein the first text data is used to describe the execution process of the target task performed by the question-answering processing model; the conversion module 1304 is used to perform language conversion on the first text data using a text conversion model to obtain second text data in the target language, wherein the target language is used to represent the language corresponding to the domain to which the target task belongs; and the generation module 1306 is used to generate target prompt words based on the second text data, wherein the target prompt words are used to guide the question-answering processing model to perform the target task.

[0168] According to an embodiment of this application, a prompt word generation apparatus for implementing the above-described prompt word generation method is also provided. Figure 14 This is a schematic diagram of another optional prompt word generation device according to an embodiment of this application, such as... Figure 14 As shown, the device includes: a calling module 1402, a conversion module 1404, a generation module 1406, and an output module 1408.

[0169] The system includes the following modules: 1402, which retrieves first text data in natural language by calling a first interface, wherein the first interface includes a first parameter whose value includes the first text data, and the first text data describes the execution process of the target task performed by the question-answering processing model; 1404, which performs language conversion on the first text data using a text conversion model to obtain second text data in the target language, wherein the target language represents the language corresponding to the domain of the target task; 1406, which generates target prompt words based on the second text data, wherein the target prompt words guide the question-answering processing model to perform the target task; and 1408, which outputs the target prompt words by calling a second interface, wherein the second interface includes a second parameter whose value includes the target prompt words.

[0170] Embodiments of this application may provide an electronic device, which may be any one of a group of electronic devices. Optionally, in this embodiment, the aforementioned electronic device may also be replaced by a terminal device such as a mobile terminal.

[0171] Optionally, in this embodiment, the aforementioned electronic device may be located in at least one of a plurality of network devices in a computer network.

[0172] In this embodiment, the electronic device described above can execute the program code in the method.

[0173] Optionally, Figure 15 This is a structural block diagram of an electronic device according to an embodiment of this application, such as... Figure 15As shown, the electronic device A may include: one or more (only one is shown in the figure) processors 1502, memory 1504, memory controller, and peripheral interfaces, wherein the peripheral interfaces are connected to the radio frequency module, the audio module, and the display.

[0174] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the methods and apparatus in the embodiments of this application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby implementing the methods in the above embodiments. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to terminal A via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0175] The processor can invoke information and application programs stored in memory via a transmission device to perform the following steps: In response to a first input instruction applied to the operating interface, displaying first text data in natural language on the operating interface, wherein the first text data describes the execution process of the target task performed by the question-answering processing model; In response to a conversion instruction applied to the operating interface, displaying a conversion window on the operating interface, wherein the conversion window displays second text data in the target language, the second text data being text data obtained by language conversion of the first text data, the target language being used to represent the language corresponding to the domain of the target task; In response to a first release instruction applied to the operating interface, displaying a first release result corresponding to the first release instruction on the operating interface, wherein the first release result is used to represent the result of generating target prompt words based on the second text data, the target prompt words being used to guide the question-answering processing model to perform the target task.

[0176] Optionally, the method further includes: assembling the first text data into a first prompt word based on a preset prompt word template; and using the first prompt word to guide the text conversion model to generate the second text data.

[0177] Optionally, the method further includes: obtaining a target prompt word template that matches the target prompt word, and an interface plugin that matches the target task, wherein the interface plugin is used to represent the interface called by the question-answering processing model during the execution of the target task; and assembling the second text data into target prompt words based on the target prompt word template and the interface plugin.

[0178] Optionally, after displaying the conversion window on the operation interface, the method further includes: in response to a confirmation command applied to the conversion window, displaying an original flowchart on the operation interface, wherein the original flowchart is a flowchart obtained by converting the second text data, wherein the original flowchart contains multiple process nodes and edges between different process nodes, wherein different process nodes are used to represent different execution steps of the target task, and edges are used to represent the execution order between different process nodes; and in response to a first release command, displaying a first release result on the operation interface, wherein the first release result is used to represent the result of generating target prompt words based on the original flowchart.

[0179] Optionally, the method further includes: responding to a modification instruction applied to the operation interface, displaying a target flowchart on the operation interface, wherein the target flowchart is a flowchart obtained by modifying the original flowchart based on the modification instruction; responding to a second release instruction applied to the operation interface, displaying a second release result corresponding to the second release instruction on the operation interface, wherein the second release result is used to characterize the result of generating target prompt words based on third text data, and the third text data is text data obtained by converting the target flowchart into a target language.

[0180] Optionally, before responding to the first release command applied to the operation interface, the method further includes: responding to the debugging command applied to the operation interface by displaying a debugging window on the operation interface; responding to the second input command applied to the debugging window by displaying query text data corresponding to the second input command in the debugging window; and displaying response text data corresponding to the query text data in the debugging window, wherein the response text data is text data generated by using the second text data to guide the question-answering processing model to process the query text data.

[0181] Optionally, the method further includes: parsing the second text data to determine multiple process nodes contained in the second text data; determining prompt word templates corresponding to the multiple process nodes based on the node types of the multiple process nodes; assembling the second text data into prompt words corresponding to the multiple process nodes based on the prompt word templates corresponding to the multiple process nodes; and sequentially using the prompt words corresponding to the multiple process nodes to guide the question-answering processing model to process the query text data to obtain the response text data.

[0182] Optionally, the question-and-answer processing model is guided to process the query text data and obtain the response text data by sequentially using prompts corresponding to multiple process nodes. This includes: determining the target process node to be executed among multiple process nodes; using prompts corresponding to the target process node to guide the question-and-answer processing model to process historical processing results and obtain the target processing result corresponding to the target process node, wherein, if the target process node is the first executed process node among multiple process nodes, the historical processing result is the query text data; if the target process node is not the first executed process node, the historical processing result is the processing result corresponding to the process node preceding the target process node; if the target process node is not the last executed process node among multiple process nodes, the process node following the target process node is taken as the target process node, and the step of guiding the question-and-answer processing model to process historical processing results and obtain the target processing result by using prompts corresponding to the target process node is repeated until the target process node is the last executed process node; if the target process node is the last executed process node among multiple process nodes, the target processing result is determined to be the response text data.

[0183] Optionally, in response to a viewing command applied to the debugging window, the method further includes: displaying a process execution window on the operation interface, wherein the process execution window displays the original flowchart and the execution status of different process nodes in the original process, wherein the execution status is used to characterize whether the corresponding process node has been executed successfully.

[0184] Optionally, in response to a first input instruction applied to the operation interface, the method further includes at least one of the following: in response to a template selection instruction applied to the operation interface, displaying a set of prompt word templates corresponding to the template selection instruction on the operation interface, wherein different prompt word templates in the template set are used to assemble different text data into prompt words; in response to a model selection instruction applied to the operation interface, displaying the type corresponding to the question-answering processing model on the operation interface; in response to a plugin selection instruction applied to the operation interface, displaying an interface plugin that matches the target task on the operation interface.

[0185] 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, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation portals are provided for users to choose to authorize or refuse.

[0186] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0187] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, they can also be implemented by hardware. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0188] Embodiments of this application also provide a computer-readable storage medium. Optionally, in this embodiment, the computer-readable storage medium can be used to store program code executed by the following method.

[0189] In response to a first input command applied to the operation interface, the system displays first text data in natural language on the operation interface. This first text data describes the execution process of the target task performed by the question-answering processing model. In response to a conversion command applied to the operation interface, a conversion window is displayed on the operation interface. This conversion window displays second text data in the target language, which is obtained by converting the first text data into a different language. The target language represents the language corresponding to the domain of the target task. In response to a first release command applied to the operation interface, the system displays a first release result corresponding to the first release command. This first release result represents the result of generating target prompt words based on the second text data. The target prompt words guide the question-answering processing model to perform the target task.

[0190] Optionally, the method further includes: assembling the first text data into a first prompt word based on a preset prompt word template; and using the first prompt word to guide the text conversion model to generate the second text data.

[0191] Optionally, the method further includes: obtaining a target prompt word template that matches the target prompt word, and an interface plugin that matches the target task, wherein the interface plugin is used to represent the interface called by the question-answering processing model during the execution of the target task; and assembling the second text data into target prompt words based on the target prompt word template and the interface plugin.

[0192] Optionally, after displaying the conversion window on the operation interface, the method further includes: in response to a confirmation command applied to the conversion window, displaying an original flowchart on the operation interface, wherein the original flowchart is a flowchart obtained by converting the second text data, wherein the original flowchart contains multiple process nodes and edges between different process nodes, wherein different process nodes are used to represent different execution steps of the target task, and edges are used to represent the execution order between different process nodes; and in response to a first release command, displaying a first release result on the operation interface, wherein the first release result is used to represent the result of generating target prompt words based on the original flowchart.

[0193] Optionally, the method further includes: responding to a modification instruction applied to the operation interface, displaying a target flowchart on the operation interface, wherein the target flowchart is a flowchart obtained by modifying the original flowchart based on the modification instruction; responding to a second release instruction applied to the operation interface, displaying a second release result corresponding to the second release instruction on the operation interface, wherein the second release result is used to characterize the result of generating target prompt words based on third text data, and the third text data is text data obtained by converting the target flowchart into a target language.

[0194] Optionally, before responding to the first release command applied to the operation interface, the method further includes: responding to the debugging command applied to the operation interface by displaying a debugging window on the operation interface; responding to the second input command applied to the debugging window by displaying query text data corresponding to the second input command in the debugging window; and displaying response text data corresponding to the query text data in the debugging window, wherein the response text data is text data generated by using the second text data to guide the question-answering processing model to process the query text data.

[0195] Optionally, the method further includes: parsing the second text data to determine multiple process nodes contained in the second text data; determining prompt word templates corresponding to the multiple process nodes based on the node types of the multiple process nodes; assembling the second text data into prompt words corresponding to the multiple process nodes based on the prompt word templates corresponding to the multiple process nodes; and sequentially using the prompt words corresponding to the multiple process nodes to guide the question-answering processing model to process the query text data to obtain the response text data.

[0196] Optionally, the question-and-answer processing model is guided to process the query text data and obtain the response text data by sequentially using prompts corresponding to multiple process nodes. This includes: determining the target process node to be executed among multiple process nodes; using prompts corresponding to the target process node to guide the question-and-answer processing model to process historical processing results and obtain the target processing result corresponding to the target process node, wherein, if the target process node is the first executed process node among multiple process nodes, the historical processing result is the query text data; if the target process node is not the first executed process node, the historical processing result is the processing result corresponding to the process node preceding the target process node; if the target process node is not the last executed process node among multiple process nodes, the process node following the target process node is taken as the target process node, and the step of guiding the question-and-answer processing model to process historical processing results and obtain the target processing result by using prompts corresponding to the target process node is repeated until the target process node is the last executed process node; if the target process node is the last executed process node among multiple process nodes, the target processing result is determined to be the response text data.

[0197] Optionally, in response to a viewing command applied to the debugging window, the method further includes: displaying a process execution window on the operation interface, wherein the process execution window displays the original flowchart and the execution status of different process nodes in the original process, wherein the execution status is used to characterize whether the corresponding process node has been executed successfully.

[0198] Optionally, in response to a first input instruction applied to the operation interface, the method further includes at least one of the following: in response to a template selection instruction applied to the operation interface, displaying a set of prompt word templates corresponding to the template selection instruction on the operation interface, wherein different prompt word templates in the template set are used to assemble different text data into prompt words; in response to a model selection instruction applied to the operation interface, displaying the type corresponding to the question-answering processing model on the operation interface; in response to a plugin selection instruction applied to the operation interface, displaying an interface plugin that matches the target task on the operation interface.

[0199] Embodiments of this application also provide a computer program product. Optionally, in this embodiment, the computer program product may include a computer program that, when executed by a processor, implements the methods provided in the embodiments described above.

[0200] The aforementioned computer program products can refer to software programs that have been written, tested, and released, and can run on computers or other devices. Computer program products can include application programs, operating systems, utility software, etc., used to achieve specific functions or solve specific problems.

[0201] Embodiments of this application also provide a computer program product. Optionally, the computer program product may include a non-volatile computer-readable storage medium, which can be used to store a computer program that, when executed by a processor, implements the method provided in the above embodiments.

[0202] The aforementioned non-volatile computer-readable storage medium can refer to a medium for storing data. Non-volatile computer-readable storage media can retain data without loss when power is off and can be used to store long-term data, such as operating systems, applications, and user files. Non-volatile storage media can include hard disk drives, solid-state drives, optical disks, and flash memory storage devices, etc.

[0203] Embodiments of this application also provide a computer program. Optionally, in this embodiment, when the computer program is executed by a processor, it implements the method provided in the above embodiments.

[0204] The aforementioned computer program can refer to a set of instructions used to tell the computer to perform specific tasks or operations. Computer programs can be written by programmers using specific programming languages ​​and can include algorithms, data structures, logic, and control flow. Computer programs can be used for a variety of purposes, including application software, operating systems, etc.

[0205] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0206] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0207] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0208] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0209] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0210] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for generating prompt words, characterized in that, include: In response to a first input command applied to the operation interface, first text data in natural language is displayed on the operation interface, wherein the first text data is used to describe the execution process of the target task performed by the question-answering processing model; In response to a conversion command applied to the operation interface, a conversion window is displayed on the operation interface, wherein the conversion window displays second text data in the target language, the second text data being text data obtained by language conversion of the first text data, and the target language being used to represent the language corresponding to the domain to which the target task belongs; In response to a first release command applied to the operation interface, a first release result corresponding to the first release command is displayed on the operation interface, wherein the first release result is used to characterize the result of generating target prompt words based on the second text data, and the target prompt words are used to guide the question-answering processing model to perform the target task.

2. The method according to claim 1, characterized in that, The method further includes: Based on a preset prompt word template, the first text data is assembled into a first prompt word; The first prompt word guides the text conversion model to generate the second text data.

3. The method according to claim 1, characterized in that, The method further includes: Obtain a target prompt word template that matches the target prompt word, and an interface plugin that matches the target task, wherein the interface plugin is used to represent the interface called by the question answering processing model during the execution of the target task; Based on the target prompt word template and the interface plugin, the second text data is assembled into the target prompt word.

4. The method according to any one of claims 1 to 3, characterized in that, After the conversion window is displayed on the user interface, the method further includes: In response to a confirmation command applied to the conversion window, the original flowchart is displayed on the operation interface. The original flowchart is a flowchart obtained by converting the second text data. The original flowchart contains multiple process nodes and edges between different process nodes. Different process nodes are used to represent different execution steps of the target task, and the edges are used to represent the execution order between the different process nodes. In response to the first release instruction, the first release result is displayed on the operation interface, wherein the first release result is used to characterize the result of generating the target prompt word based on the original flowchart.

5. The method according to claim 4, characterized in that, The method further includes: In response to a modification command applied to the operation interface, a target flowchart is displayed on the operation interface, wherein the target flowchart is a flowchart obtained by modifying the original flowchart based on the modification command; In response to a second release command applied to the operation interface, a second release result corresponding to the second release command is displayed on the operation interface. The second release result is used to characterize the result of generating the target prompt word based on third text data, which is text data obtained by converting the target flowchart into the target language.

6. The method according to any one of claims 1 to 3, characterized in that, Prior to responding to a first release command applied to the user interface, the method further includes: In response to a debugging command applied to the operation interface, a debugging window is displayed on the operation interface; In response to a second input command applied to the debug window, the query text data corresponding to the second input command is displayed in the debug window; The debug window displays the response text data corresponding to the query text data, wherein the response text data is generated by using the second text data to guide the question-and-answer processing model to process the query text data.

7. The method according to claim 6, characterized in that, The method further includes: The second text data is parsed to determine the multiple process nodes contained in the second text data; Based on the node types of the multiple process nodes, determine the prompt word templates corresponding to the multiple process nodes; Based on the prompt word templates corresponding to the multiple process nodes, the second text data is assembled into prompt words corresponding to the multiple process nodes; The prompts corresponding to the multiple process nodes are used sequentially to guide the question-and-answer processing model to process the query text data and obtain the response text data.

8. The method according to claim 7, characterized in that, The step of sequentially using prompts corresponding to the multiple process nodes to guide the question-and-answer processing model to process the query text data and obtain the response text data includes: Determine the target process node to be executed among the plurality of process nodes; Using the prompt words corresponding to the target process node, the question-answering processing model is guided to process the historical processing results to obtain the target processing result corresponding to the target process node. Wherein, if the target process node is the first process node executed among the plurality of process nodes, the historical processing result is the query text data. If the target process node is not the first process node executed, the historical processing result is the processing result corresponding to the process node executed before the target process node. If the target process node is not the last executed process node among the plurality of process nodes, the next executed process node after the target process node is taken as the target process node, and the step of using the prompt words corresponding to the target process node to guide the question-answering processing model to process the historical processing results and obtain the target processing result is repeated until the target process node is the last executed process node. If the target process node is the last process node executed among the plurality of process nodes, the target processing result is determined to be the response text data.

9. The method according to claim 6, characterized in that, In response to a viewing command applied to the debug window, the method further includes: The operation interface displays a process execution window, which shows the original flowchart and the execution status of different process nodes in the original flowchart. The execution status is used to indicate whether the corresponding process node has been executed successfully.

10. The method according to any one of claims 1 to 3, characterized in that, In response to a first input command applied to the user interface, the method further includes at least one of the following: In response to a template selection command applied to the operation interface, a set of prompt word templates corresponding to the template selection command is displayed on the operation interface, wherein different prompt word templates in the template set are used to assemble different text data into prompt words; In response to a model selection command applied to the operation interface, the type corresponding to the question-and-answer processing model is displayed on the operation interface; In response to a plugin selection command applied to the operation interface, an interface plugin matching the target task is displayed on the operation interface.

11. A text processing method, characterized in that, include: In response to an input command applied to the interactive interface, the current query text is displayed on the interactive interface; In response to a processing instruction applied to the interactive interface, the current response text corresponding to the current query text is displayed on the interactive interface, wherein the current response text is generated by a question-and-answer processing model that uses a target prompt word to guide the processing of the current query text, and the target prompt word is a prompt word generated by the method described in any one of claims 1 to 7.

12. A method for generating prompt words, characterized in that, include: Acquire first text data in natural language, wherein the first text data is used to describe the execution process of the target task performed by the question answering processing model; The first text data is converted to a second text data in the target language, wherein the target language is used to represent the language corresponding to the domain to which the target task belongs; Target prompts are generated based on the second text data, wherein the target prompts are used to guide the question-answering processing model to perform the target task.

13. A method for generating prompt words, characterized in that, include: The first text data of natural language is obtained by calling the first interface, wherein the first interface includes a first parameter, the parameter value of the first parameter includes the first text data, and the first text data is used to describe the execution process of the target task executed by the question answering processing model. The first text data is converted to a second text data in the target language, wherein the target language is used to represent the language corresponding to the domain to which the target task belongs; Target prompt words are generated based on the second text data, wherein the target prompt words are used to guide the question-answering processing model to perform the target task; The target prompt word is output by calling a second interface, wherein the second interface includes a second parameter, and the parameter value of the second parameter includes the target prompt word.

14. An electronic device, characterized in that, include: Memory, which stores executable programs; A processor for running the program, wherein the program, when running, performs the method according to any one of claims 1 to 13.

15. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored executable program, wherein, when the executable program is executed, it controls the device on which the computer-readable storage medium is located to perform the method according to any one of claims 1 to 13.

16. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method described in any one of claims 1 to 13.

Images