Implementation method for embedding business process call into dialogue system by using large language model
Patent Information
- Application Number
- CN202311431408.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-10-31
- Publication Date
- 2026-11-03
- Estimated Expiration
- 2043-10-31
AI Technical Summary
Existing technologies cannot intelligently switch between question-and-answer sessions and business flow calls during the question-and-answer process, nor can they combine historical question-and-answer information to identify user intent and automatically determine the business processes that need to be called.
The system encapsulates user input and historical Q&A information using a large language model. It then processes this information through Plugin Prompt and QA Prompt to determine whether a business process needs to be invoked. Based on the process type, it decides whether to directly invoke or repeatedly invoke the business process.
It enables intelligent switching between question-and-answer and business flow invocation in the dialogue system, automatically identifies and invokes the appropriate business process, and improves the intelligence and automation level of the question-and-answer system.
Smart Images
Figure CN117371539B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of general intelligent question answering technology, specifically to a method for embedding business process calls into a dialogue system using a large language model. Background Technology
[0002] Intelligent question-answering assistants are software programs based on artificial intelligence technology that can intelligently provide accurate answers or solutions based on the user's questions or needs. Through technologies such as deep learning and natural language processing, intelligent question-answering assistants can understand and analyze user questions and provide corresponding answers. Intelligent question-answering assistants play an increasingly important role in today's society. As a significant application of artificial intelligence technology, intelligent question-answering assistants have brought tremendous convenience to people's lives and work.
[0003] Analyzing existing technologies reveals two main issues: firstly, they cannot intelligently switch between question-and-answer sessions and business flow calls during the question-and-answer process; secondly, they cannot combine historical question-and-answer information to identify user intent and thus automatically determine the business processes that need to be invoked.
[0004] Furthermore, the background technology also involves large language models. Large language models (LLMs) refer to deep learning models trained using large amounts of text data that can generate natural language text or understand the meaning of language text. Specifically, large language models are language models with a large number of parameters, similar to ChatGPT. Large language models can handle various natural language tasks, such as text classification, question answering, and dialogue, and are an important pathway to artificial intelligence. Summary of the Invention
[0005] This application provides a method for embedding business process calls into a dialogue system using a large language model, in order to solve the aforementioned technical problems in the prior art.
[0006] According to one aspect of this application, one embodiment provides a method for embedding business process calls into a dialogue system using a large language model, comprising the following steps:
[0007] S1. Encapsulate the user's current input information and historical question-and-answer information using the first method, and input them into the large language model;
[0008] S2. Determine whether a business flow needs to be invoked; if yes, proceed to step S4; if no, proceed to step S3.
[0009] S3. The user's current input information and historical question-and-answer information are encapsulated using the second method, and input into the large language model, and the return result of the large language model is output to the dialogue system;
[0010] S4. Determine whether the business process to be invoked is a one-way process; if yes, directly invoke the business process, return the result and output it to the dialogue system; if no, proceed to step S5.
[0011] S5. Encapsulate the user's current input information, historical question-and-answer information, and current call result using the first method, and input them into the large language model to determine whether the business flow needs to be called again; if so, repeat step S4 until the termination condition is met; if not, return the result and output it to the dialogue system.
[0012] In one embodiment, the first packaging method is Plugin Prompt, and the second packaging method is QAPrompt.
[0013] In one embodiment, Plugin Prompt is used to encapsulate Q1, A1, Q2, A2, ..., Qn; where Q1 represents the user's first query, i.e., the question; A1 represents the response to Q1; and Qn represents the user's nth query.
[0014] In one embodiment, in step S4, the one-way process refers to the process where the user's interaction after initiation will exit the dialogue system.
[0015] In one embodiment, in step S4, the determination of the one-way process is obtained from the constructed business process type information table.
[0016] In one embodiment, the dialogue round ends after the result is returned and output to the dialogue system.
[0017] In one embodiment, the large language model includes a deep learning model based on the Transformers Decoder architecture.
[0018] According to one aspect of this application, one embodiment provides an implementation apparatus for embedding business process calls into a dialogue system using a large language model, comprising:
[0019] The input module is used to execute step S1; step S1 is to encapsulate the user's current input information and historical question-and-answer information in the first method and input them into the large language model.
[0020] The first judgment module is used to execute step S2; step S2 is to determine whether it is necessary to call the business flow; if yes, execute step S4; if no, execute step S3.
[0021] The first processing module is used to execute step S3; step S3 is to encapsulate the user's current input information and the historical question-and-answer information in the second way, input them into the large language model, and output the return result of the large language model to the dialogue system.
[0022] The second judgment module is used to execute step S4; step S4 is to determine whether the business process to be invoked is a one-way process; if so, the business process is directly invoked, the result is returned and output to the dialogue system; if not, step S5 is executed; and
[0023] The second processing module is used to execute step S5. Step S5 is to encapsulate the user's current input information, the historical question and answer information, and the current call result in the first method, and input them into the large language model to determine whether the business flow needs to be called again. If yes, then step S4 is executed again until the termination condition is met. If no, the result is returned and output to the dialogue system.
[0024] According to one aspect of this application, one embodiment provides an electronic device, including:
[0025] Memory; and
[0026] processor;
[0027] The memory is used to store one or more computer instructions; the one or more computer instructions are executed by the processor to implement the implementation method of embedding business process calls into a dialogue system using a large language model as described above.
[0028] According to one aspect of this application, one embodiment provides a readable storage medium storing computer instructions; wherein, when the computer instructions are executed by a processor, they implement the implementation method of embedding business process calls into a dialogue system using a large language model as described in any of the above claims.
[0029] The embodiments described above in this application can determine whether a business process needs to be invoked and the appropriate process type based on the user's current input and historical question-and-answer information within a dialogue system. Specifically, this technical solution embeds business process invocation into the dialogue system, enabling intelligent switching between question-and-answer and business flow invocation without additional intervention; it can automatically identify the business process to be invoked by comprehensively considering the user's current question and historical question-and-answer information. Attached Figure Description
[0030] Figure 1 This is a flowchart illustrating how a large language model is used to embed business flow calls into a dialogue system in one embodiment.
[0031] Figure 2 This is a schematic diagram of the structure of an implementation device that uses a large language model to embed business process calls into a dialogue system in one embodiment;
[0032] Figure 3 This is a schematic diagram of a dialogue system in one embodiment. Detailed Implementation
[0033] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.
[0034] 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 merely some, not all, of the embodiments of the present application. All other embodiments obtained by those skilled in the art based on the embodiments of the present application without creative effort should fall within the scope of protection of the present application.
[0035] It should be noted that the terms "first," "second," etc., used in 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 used interchangeably where appropriate for the embodiments of this application 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 steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0036] It should be understood that when an element (such as a layer, film, region, or substrate) is described as being "on" another element, the element may be directly on the other element, or there may be an intermediate element present. Moreover, in this application, when an element is described as being "connected" to another element, the element may be "directly connected" to the other element, or "connected" to the other element via a third element.
[0037] Example 1
[0038] Please refer to Figure 1 One embodiment provides a method for embedding business process calls into a dialogue system using a large language model, including the following steps:
[0039] S1. Encapsulate the user's current input information and historical question-and-answer information using the first method, and input them into the large language model; the first encapsulation method here is Plugin Prompt; the following is an example, using Plugin Prompt to encapsulate Q1, A1, Q2, A2, ..., Qn; where Q1 represents the user's first query, i.e., the question; A1 represents the response to Q1; Qn represents the user's nth query;
[0040] S2. Determine whether a business flow needs to be invoked; if yes, proceed to step S4; if no, proceed to step S3.
[0041] S3. The user's current input information and historical question-and-answer information are encapsulated using the second method, and input into the large language model, and the return result of the large language model is output to the dialogue system; the second encapsulation method here is QA Prompt;
[0042] S4. Determine whether the business process to be invoked is a one-way process; if yes, directly invoke the business process, return the result and output it to the dialogue system; if no, proceed to step S5.
[0043] S5. Encapsulate the user's current input information, historical question-and-answer information, and current call result using the first method, and input them into the large language model to determine whether the business flow needs to be called again; if so, repeat step S4 until the termination condition is met; if not, return the result and output it to the dialogue system.
[0044] In one embodiment, in step S4, the one-way process refers to a process where the user's interaction after initiation will exit the dialogue system. Furthermore, the determination of the one-way process is obtained from the constructed business process type information table.
[0045] In one embodiment, the dialogue round ends after the result is returned and output to the dialogue system.
[0046] In one embodiment, the large language model includes a deep learning model based on the Transformers Decoder architecture; the deep learning model is such as ChatGLM, LLaMA, etc.
[0047] The present invention provides a method for embedding business process calls into a dialogue system using a large language model. This method enables the dialogue system to automatically determine whether a business process needs to be called and the appropriate process type based on the user's current input and historical question-and-answer information. The overall process is as follows: Figure 1 As shown.
[0048] Example 2
[0049] Please refer to Figure 1 One embodiment provides a method for embedding business process calls into a dialogue system using a large language model, including the following steps:
[0050] S1: The user's current input information is Qn, and the historical question-and-answer information is Q1, A1, ..., Qn-1, An-1. PluginPrompt is used to encapsulate Q1, A1, ..., Qn and input into the large language model. For PluginPrompt, please refer to:
[0051] <|START-SYS|>\nTry to answer user questions as much as possible. You can use these business flow types:\nBusiness Flow 1: Functional description of Business Flow 1. Parameters:[{\"method\":\"Method Name\",\"description\":\"Method Description\",\"required\":true,\"schema\":{\"type\":\"string\"}}]Format the arguments as a JSON object.\nBusiness Flow 2: Functional description of Business Flow 1. Parameters:[{\"method\":\"Method Name\",\"description\":\"Method Description\",\"required\":true,\"schema\":{\"type\":\"string\"}}]Format the arguments as a JSON object.\n...\nBusiness Flow N: Functional description of Business Flow N. Parameters: [{"method":"method name","description":"method description","required":true,"schema":{"type":"string"}}] Format the arguments as a JSON object. Default: Use when you can answer user questions without external plugins.Parameters: [{\"name\":\"empty_string\",\"description\":\"Empty string\",\"required\":false,\"schema\":{\"type\":\"string\"}}] Format the arguments as a JSON object.\n\nPlease answer in the following format:\n\n<|Question|>You must answer the question you entered\n<|Thinking|>You should keep thinking about how to solve the problem\n<|Plug-in|>You should take one of [Business Process 1, Business Process 2, ..., Business Process N, default]\n<|Plug-in Input|>The input of the plug-in\n<|Plug-in Output|>The output of the plug-in\n...(<|Thinking|> / <|Plug-in|> / <|Plug-in Input|> / <|Plug-in Output|>) can be repeated multiple times\n<|Thinking|>I know the final result\n<|Final Result|>Output the final result for the original question\n<|END-SYS|>\n\nStart! <|Question|> Question: Q1\nAnswer: A1\n,...,Question: Qn\n<|Thinking|>.
[0052] S2: If the output of the large language model in step S1 is Yes, proceed to step S4; otherwise, proceed to step S3.
[0053] S3: Encapsulate Q1, A1, ..., Qn using the QA Prompt and input it into the large language model. Return the results from the large language model to the dialogue system, ending this round of dialogue. For the Plugin Prompt, please refer to:
[0054] <|START-SYS|>\nYou are an intelligent assistant named <|model|>.\n-<|model|> can fluently understand and communicate based on the user's language (e.g., Chinese), and you can perform any language-based task.\n-If <|model|> does not know the answer to a question, please do not answer randomly. You can say, "I don't have an idea about this question at the moment."\n-If the user corrects an incorrect answer, <|model|> will apologize and accept the user's suggestion.\n<|END-SYS|>\n\nQuestion: Q1\nAnswer: A1\n...\nQuestion: Qn\nAnswer:
[0055] S4: Determine if the business process to be invoked is a one-way process. If yes, directly invoke the business process and return An to the dialogue system, ending this round of dialogue; otherwise, proceed to step S5. A one-way process generally refers to a process where the user's interaction will exit the dialogue system after invocation, such as invoice reimbursement. The determination of a one-way process can be obtained by querying the constructed business process type information table.
[0056] Table 1 Business Process Information Table
[0057]
[0058] S5: Initiate the business flow and obtain the call result. Then, use Plugin Prompt to encapsulate Q1, A1, ..., Qn and the current call result, and input it into the large language model to determine whether the business flow needs to be called again. If the large language model returns that the business flow needs to be called again, jump to step S4. Otherwise, return to An to the dialogue system and end this round of dialogue.
[0059] Among them, large language models include, but are not limited to, deep learning models based on the Transformers Decoder structure, such as ChatGLM and the LLaMA series.
[0060] Example 3
[0061] Please refer to Figure 2 Based on the same inventive concept, one embodiment of this application provides an implementation device for embedding business process calls into a dialogue system using a large language model, in the following form:
[0062] 1. Input module
[0063] The input module 10 is used to execute step S1; step S1 is to encapsulate the user's current input information and historical question-and-answer information in the first way and input them into the large language model.
[0064] 2. First Judgment Module
[0065] The first judgment module 20 is used to execute step S2; step S2 is to determine whether it is necessary to call the business flow; if yes, execute step S4; if no, execute step S3.
[0066] 3. First processing module
[0067] The first processing module 30 is used to execute step S3; step S3 is to encapsulate the user's current input information and the historical question-and-answer information in the second way, input them into the large language model, and output the return result of the large language model to the dialogue system.
[0068] 4. Second Judgment Module
[0069] The second judgment module 40 is used to execute step S4; step S4 is to determine whether the business process type to be called is a one-way process; if yes, the business process is directly invoked, the result is returned and output to the dialogue system; if no, step S5 is executed.
[0070] 5. Second processing module
[0071] The second processing module 50 is used to execute step S5; step S5 is to encapsulate the user's current input information, the historical question and answer information and the current call result in the first method, and input them into the large language model to determine whether the business flow needs to be called again; if so, step S4 is executed again until the termination condition is met; if not, the result is returned and output to the dialogue system.
[0072] In this embodiment, each module of the device for embedding business process calls into a dialogue system using a large language model has a one-to-one correspondence with each step of the method for embedding business process calls into a dialogue system using a large language model in the above embodiments, and will not be described again.
[0073] Example 4
[0074] Based on the same inventive concept, one embodiment of this application provides an electronic device, including: a memory and a processor; wherein, the memory is used to store one or more computer instructions; the one or more computer instructions are executed by the processor using any of the above embodiments to implement the device for embedding business process calls into a dialogue system using a large language model.
[0075] One or more of the aforementioned computer instructions can form a program.
[0076] Example 5
[0077] Based on the same inventive concept, one embodiment of this application provides a readable storage medium storing computer instructions; wherein, when the computer instructions are executed by a processor, they implement the implementation apparatus for embedding business process calls into a dialogue system using a large language model as described in any of the above embodiments.
[0078] One or more of the aforementioned computer instructions can form a program.
[0079] The aforementioned program can run on a processor or be stored in memory (or computer-readable medium). Computer-readable medium includes both permanent and non-permanent, removable and non-removable media, and information storage can be achieved by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable medium does not include transient computer-readable media, such as modulated data signals and carrier waves.
[0080] These computer programs may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes can be implemented using different modules, and different steps can be implemented using different modules.
[0081] Example 6
[0082] Please refer to Figure 3 One embodiment provides a specific application of a method for embedding business process calls into a dialogue system using a large language model. The method for embedding business process calls into a dialogue system using a large language model can be selected from the above embodiment.
[0083] Step 1
[0084] Input of large language models
[0085] <|START-SYS|>
[0086] To answer user questions as thoroughly as possible, you can use these plugins:
[0087] `calculate`: This tool is used to interact with the Calculator API. The Calculator API can be used to answer mathematical calculations such as addition, subtraction, multiplication, and division. Parameters: [{"name":"expression","description":"Mathematical expression connected by +-* / symbols, used for numerical calculations","required":true,"schema":{"type":"string"}}] Format the arguments as a JSON object.
[0088] `reimburse`: This tool is used to interact with the expense reimbursement API. The expense reimbursement API helps with invoice reimbursement. Parameters: [{"name":"bill_id","description":"A string of letters and numbers describing the expense document","required":true,"schema":{"type":"string"}}] Format the arguments as a JSON object.
[0089] default: Use this when you can answer user questions without external plugins. Parameters: [{"name":"empty_string","description":"Empty string","required":false,"schema":{"type":"string"}}] Format the arguments as a JSON object.
[0090] Please answer in the following format:
[0091] <|Question|> You must answer the question you entered.
[0092] <|Thinking|> You should always keep thinking, thinking about how to solve problems.
[0093] <|Plugin|> You should take one of [calculate, reimburse, default]
[0094] <|Plug-in Input|>Plug-in Input
[0095] <|Plugin Output|> Plugin Output
[0096] ...(<|Thinking|> / <|Plug-in|> / <|Plug-in Input|> / <|Plug-in Output|>) can be repeated multiple times.
[0097] <|Thinking|> I know the final result!
[0098] <|Final Result|> Output the final result for the original problem.
[0099] <|END-SYS|>
[0100] start!
[0101] <|Question|>
[0102] Q: Please introduce China.
[0103] A: China is an ancient civilization with a long history and a population of over 1.4 billion, making it the world's largest developing country. China possesses abundant natural resources, a rich historical and cultural heritage, numerous famous tourist attractions, renowned cuisines, rich cultural traditions, and many famous historical figures.
[0104] Q: Hello, I need an invoice with reimbursement code KN38947980.
[0105] A: The reimbursement process for invoice number KN38947980 has been initiated.
[0106] Q: Mina had 8 pears, then bought 5 bananas. Mina ate 1 pear. How much fruit did Mina have left?
[0107] <|Thinking|>
[0108] Output of large language model
[0109] How much fruit does Mina have left?
[0110] <|Plugin|>calculate
[0111] <|Plugin Input|>{'input':'8+5-1='}
[0112] Step 5
[0113] Output of business flow 12
[0115] Recall the input of the large language model
[0116] <|START-SYS|>
[0117] To answer user questions as thoroughly as possible, you can use these plugins:
[0118] `calculate`: This tool is used to interact with the Calculator API. The Calculator API can be used to answer mathematical calculations such as addition, subtraction, multiplication, and division. Parameters: [{"name":"expression","description":"Mathematical expression connected by +-* / symbols, used for numerical calculations","required":true,"schema":{"type":"string"}}] Format the arguments as a JSON object.
[0119] `reimburse`: This tool is used to interact with the expense reimbursement API. The expense reimbursement API helps with invoice reimbursement. Parameters: [{"name":"bill_id","description":"A string of letters and numbers describing the expense document","required":true,"schema":{"type":"string"}}] Format the arguments as a JSON object.
[0120] default: Use this when you can answer user questions without external plugins. Parameters: [{"name":"empty_string","description":"Empty string","required":false,"schema":{"type":"string"}}] Format the arguments as a JSON object.
[0121] Please answer in the following format:
[0122] <|Question|> You must answer the question you entered.
[0123] <|Thinking|> You should always keep thinking, thinking about how to solve problems.
[0124] <|Plugin|> You should take one of [calculate, reimburse, default]
[0125] <|Plug-in Input|>Plug-in Input
[0126] <|Plugin Output|> Plugin Output
[0127] ...(<|Thinking|> / <|Plug-in|> / <|Plug-in Input|> / <|Plug-in Output|>) can be repeated multiple times.
[0128] <|Thinking|> I know the final result!
[0129] <|Final Result|> Output the final result for the original problem.
[0130] <|END-SYS|>
[0131] start!
[0132] <|Question|>
[0133] Q: Please introduce China.
[0134] A: China is an ancient civilization with a long history and a population of over 1.4 billion, making it the world's largest developing country. China possesses abundant natural resources, a rich historical and cultural heritage, numerous famous tourist attractions, renowned cuisines, rich cultural traditions, and many famous historical figures.
[0135] Q: Hello, I need an invoice with reimbursement code KN38947980.
[0136] A: The reimbursement process for invoice number KN38947980 has been initiated.
[0137] Q: Mina had 8 pears, then bought 5 bananas. Mina ate 1 pear. How much fruit did Mina have left?
[0138] <|Thinking|> How much fruit does Mina have left?
[0139] <|Plugin|>calculate
[0140] <|Plugin Input|>{'input':'8+5-1='}
[0141] <|Plugin Output|>12
[0142] <|Thinking|>
[0143] Output of large language model
[0144] I know the final result.
[0145] <|Final Result|> Mina has 12 more fruits.
[0146] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.
Claims
1. A method for embedding business process calls into a dialogue system using a large language model, characterized in that, Includes the following steps: S1. Encapsulate the user's current input information and historical question-and-answer information using the first method, and input them into the large language model; S2. Determine whether the business logic needs to be invoked; If yes, proceed to step S4; if no, proceed to step S3. S3. The user's current input information and historical question-and-answer information are encapsulated using the second method, and input into the large language model, and the return result of the large language model is output to the dialogue system; S4. Determine whether the business process to be invoked is a one-way process; if yes, directly invoke the business process, return the result and output it to the dialogue system; if no, proceed to step S5. S5. Initiate the business flow and obtain the call result; encapsulate the user's current input information, the historical question and answer information, and the current call result using the first method, and input them into the large language model to determine whether it is necessary to call the business flow again; If yes, then repeat step S4 until the termination condition is met; if no, return the result and output it to the dialogue system. In step S4, the one-way process refers to the process where the user's interaction after initiation will exit the dialogue system.
2. The method for embedding business process calls into a dialogue system using a large language model according to claim 1, characterized in that, The first packaging method is Plugin Prompt, and the second packaging method is QA Prompt.
3. The method for embedding business process calls into a dialogue system using a large language model according to claim 1, characterized in that, The Plugin Prompt encapsulates Q1, A1, Q2, A2, ..., Qn; where Q1 represents the user's first query, i.e., the question; A1 represents the response to Q1; and Qn represents the user's nth query.
4. The method for embedding business process calls into a dialogue system using a large language model according to claim 1, characterized in that, In step S4, the determination of the one-way process is obtained from the constructed business process type information table.
5. The method for embedding business process calls into a dialogue system using a large language model according to claim 1, characterized in that, After returning the results and outputting them to the dialogue system, the current dialogue round ends.
6. A method for embedding business process calls into a dialogue system using a large language model, as described in any one of claims 1-5, characterized in that... The large language model includes a deep learning model based on the Transformers Decoder structure.
7. An implementation device for embedding business process calls into a dialogue system using a large language model, characterized in that, include: The input module is used to execute step S1; step S1 is to encapsulate the user's current input information and historical question-and-answer information in the first method and input them into the large language model. The first judgment module is used to execute step S2; step S2 is to determine whether it is necessary to call the business flow. If yes, proceed to step S4; if no, proceed to step S3. The first processing module is used to execute step S3; step S3 is to encapsulate the user's current input information and the historical question-and-answer information in the second way, input them into the large language model, and output the return result of the large language model to the dialogue system. The second judgment module is used to execute step S4; step S4 is to determine whether the business process type to be called is a one-way process; if yes, the business process is directly invoked, the result is returned and output to the dialogue system; if no, step S5 is executed. and The second processing module is used to execute step S5; step S5 is to invoke the business flow and obtain the call result; encapsulate the user's current input information, the historical question and answer information and the current call result in the first method, and input them into the large language model to determine whether the business flow needs to be invoked again; If yes, then repeat step S4 until the termination condition is met; if no, return the result and output it to the dialogue system. In step S4, the one-way process refers to the process where the user's interaction after initiation will exit the dialogue system.
8. An electronic device, characterized in that, include: Memory; and processor; The memory is used to store one or more computer instructions; the one or more computer instructions are executed by the processor to implement the implementation method of embedding business process invocation into a dialogue system using a large language model as described in any one of claims 1 to 6.
9. A readable storage medium, characterized in that, The readable storage medium stores computer instructions; wherein, when the computer instructions are executed by a processor, they implement the method for embedding business process calls into a dialogue system using a large language model as described in any one of claims 1 to 6.