Intelligent question and answer method and system, computer equipment and storage medium

The intelligent question-answering method enhances insurance systems by using multiple prompts and search engines to address semantic understanding and complexity, improving accuracy and user experience.

CN119088935APending Publication Date: 2024-12-06CHINA PING AN PROPERTY INSURANCE CO LTD

Patent Information

Application Number
CN202411202470.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-29
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

Existing insurance question-answering systems face issues with inaccurate semantic understanding, incomplete knowledge bases, poor user experience, and insufficient ability to handle complex questions, leading to low accuracy and limited service capabilities.

Method used

An intelligent question-answering method involving multiple prompts and interactions with a large language model (LLM) and search engines to generate accurate and comprehensive answers, including designing prompts to refine user questions, search plans, and integrating search results for enhanced understanding and knowledge generation.

Benefits of technology

Improves accuracy, comprehensiveness, and user experience by leveraging LLM's professional insurance knowledge and search engines to provide detailed and relevant answers to complex questions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119088935A_ABST
    Figure CN119088935A_ABST
Patent Text Reader

Abstract

The invention discloses an intelligent question and answer method and system, computer equipment and a storage medium, and relates to the technical field of artificial intelligence, and the method comprises the steps: designing a first prompt based on a target question given by a target user, inputting the first prompt and the target question into a large language model LLM, and generating a corresponding search plan; designing a second prompt based on the search plan, and inputting the second prompt and the search plan into LLM to generate a corresponding search keyword; inputting the search keyword into a search engine to generate a corresponding search result; designing a third prompt based on the search result, and inputting the third prompt, the target problem, the search plan and the search result into LLM to generate a corresponding knowledge text; and designing a fourth prompt based on the knowledge text, and inputting the fourth prompt, the target question and the knowledge text into the LLM to generate a corresponding target answer. By means of the method, the accuracy, comprehensiveness, user experience and complex transaction processing capacity during question answering can be effectively improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an intelligent question-answering method, system, computer device, and non-volatile computer-readable storage medium. Background Art

[0002] At present, with the rapid development of information technology, the application of artificial intelligence in various fields has gradually deepened, and the insurance industry is no exception. At present, the existing insurance question-and-answer systems are generally divided into two categories: FAQ insurance question-and-answer and knowledge base insurance question-and-answer. Although these two systems have made certain progress in providing insurance consulting services to users, there are still some problems and limitations, such as:

[0003] 1. Inaccurate semantic understanding leads to inaccurate answers;

[0004] Semantic understanding is the key to accurately answering user questions in insurance question-answering systems. For example, existing FAQ insurance question-answering systems mostly use similarity matching methods, that is, selecting the most appropriate answer by comparing the text similarity between user questions and questions in the knowledge base. However, this method often ignores the deep semantics and implicit information behind user questions, resulting in low accuracy of answers.

[0005] 2. The quality of the knowledge base is not high, resulting in incomplete answers;

[0006] The knowledge base is the core component of the FAQ insurance question-and-answer system. Its quality and comprehensiveness directly affect the accuracy and comprehensiveness of the system's answers. The current insurance question-and-answer system knowledge base often has problems such as untimely information updates and limited coverage, which results in the system being unable to provide satisfactory answers when faced with some specific or novel questions. This not only affects the user experience, but also limits the system's service capabilities.

[0007] 3. The conversation process is not natural enough, resulting in poor user experience;

[0008] User experience is another important indicator to measure the success of insurance Q&A systems. Existing insurance Q&A systems are often too simple in dialogue process design and lack a natural and smooth interactive experience. When users communicate with the system, they may encounter comprehension barriers or poor communication, which greatly reduces user satisfaction and the practicality of the system.

[0009] 4. Insufficient ability to handle complex issues and multi-round conversations;

[0010] When faced with complex questions or scenarios that require multiple rounds of dialogue, existing FAQ question-answering systems often seem to be unable to cope. They can usually only handle simple questions and answers, and cannot conduct in-depth logical reasoning or continuous dialogue exchanges. For example, when dealing with insurance issues involving specific drug claims, the system may not be able to accurately determine the correlation between the drug and the accident, resulting in an inability to give reasonable claims recommendations.

[0011] Based on this, how to provide an intelligent question-answering method, system, computer device and non-volatile computer-readable storage medium to effectively improve the accuracy, comprehensiveness, user experience and ability to handle complex transactions during question-answering is a problem that urgently needs to be solved by technical personnel in this field. Summary of the invention

[0012] In view of the above-mentioned shortcomings of the prior art, the purpose of the present invention is to provide an intelligent question and answer method, system, computer device and non-volatile computer-readable storage medium that can be used in financial technology or other related fields, aiming to solve the problem of how to effectively improve the accuracy, comprehensiveness, user experience and ability to handle complex transactions when answering questions.

[0013] In order to achieve the above object, the present invention adopts the following technical solutions:

[0014] An intelligent question-answering method, comprising:

[0015] Designing a first prompt based on a target question given by a target user, inputting the first prompt and the target question into a large language model LLM, and generating a corresponding search plan;

[0016] Design a second prompt based on the search plan, input the second prompt and the search plan into the LLM, and generate corresponding search keywords;

[0017] Input the search keyword into a search engine to generate corresponding search results;

[0018] Designing a third prompt based on the search results, inputting the third prompt, the target question, the search plan, and the search results into the LLM to generate a corresponding knowledge text;

[0019] A fourth prompt is designed based on the knowledge text, and the fourth prompt, the target question, and the knowledge text are input into the LLM to generate a corresponding target answer.

[0020] In a further technical solution, the intelligent question answering method, wherein the first prompt is designed based on the target question given by the target user, the first prompt and the target question are input into the large language model LLM, and a corresponding search plan is generated, including:

[0021] Get the target question input by the target user;

[0022] Designing a first initial prompt based on the target question given by the target user, and obtaining a corresponding first prompt after debugging the first initial prompt multiple times;

[0023] The first prompt and the target question are input into a preset large language model LLM to generate a corresponding search plan.

[0024] In a further technical solution, the intelligent question-answering method, wherein the second prompt is designed based on the search plan, the second prompt and the search plan are input into the LLM, and the corresponding search keywords are generated, including:

[0025] Obtaining the search plan generated by the LLM;

[0026] Designing a second initial prompt based on the search plan, and obtaining a corresponding second prompt after multiple debugging of the second initial prompt;

[0027] The second prompt and the search plan are input into the LLM to generate corresponding search keywords.

[0028] In a further technical solution, the intelligent question-answering method, wherein the step of inputting the search keyword into a search engine to generate corresponding search results, comprises:

[0029] The search keyword is input into a preset search engine to perform a search and generate corresponding search results.

[0030] In a further technical solution, the intelligent question-answering method, wherein the third prompt is designed based on the search results, the third prompt, the target question, the search plan, and the search results are input into the LLM, and the corresponding knowledge text is generated, includes:

[0031] Obtaining the search results generated by the search engine;

[0032] Designing a third initial prompt based on the search result, and obtaining a corresponding third prompt after debugging the third initial prompt multiple times;

[0033] The third prompt, the target question, the search plan, and the search results are input into the LLM to generate a corresponding knowledge text.

[0034] In a further technical solution, the intelligent question-answering method, wherein the fourth prompt is designed based on the knowledge text, the fourth prompt, the target question, and the knowledge text are input into the LLM, and the corresponding target answer is generated, comprises:

[0035] Obtaining the knowledge text generated by the LLM;

[0036] Designing a fourth initial prompt based on the knowledge text, and obtaining a corresponding fourth prompt after multiple debugging of the fourth initial prompt;

[0037] The fourth prompt, the target question, and the knowledge text are input into the LLM to generate a corresponding target answer.

[0038] In a further technical solution, the intelligent question-answering method, wherein the fourth prompt is designed based on the knowledge text, the fourth prompt, the target question, and the knowledge text are input into the LLM, and after the corresponding target answer is generated, further comprises:

[0039] The target answer is returned to the target user.

[0040] An intelligent question-answering system, comprising:

[0041] A first generating module is used to design a first prompt based on a target question given by a target user, input the first prompt and the target question into a large language model LLM, and generate a corresponding search plan;

[0042] A second generating module, configured to design a second prompt based on the search plan, input the second prompt and the search plan into the LLM, and generate corresponding search keywords;

[0043] A search module, used to input the search keyword into a search engine to generate corresponding search results;

[0044] A third generation module is used to design a third prompt based on the search results, input the third prompt, the target question, the search plan, and the search results into the LLM, and generate a corresponding knowledge text;

[0045] The fourth generation module is used to design a fourth prompt based on the knowledge text, input the fourth prompt, the target question, and the knowledge text into the LLM, and generate a corresponding target answer.

[0046] A computer device, wherein the computer device comprises at least one processor; and

[0047] a memory communicatively connected to the at least one processor; wherein,

[0048] The memory stores a computer program that can be executed by the at least one processor. When the computer program is executed by the at least one processor, the intelligent question and answer method as described in any one of the above items can be implemented.

[0049] A non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by at least one processor, it can implement the intelligent question-answering method as described in any one of the above items.

[0050] Compared with the prior art, the present invention provides an intelligent question-answering method, system, computer device and non-volatile computer-readable storage medium, wherein the method includes: designing a first prompt based on a target question given by a target user, inputting the first prompt and the target question into a large language model LLM, and generating a corresponding search plan; designing a second prompt based on the search plan, inputting the second prompt and the search plan into the LLM, and generating corresponding search keywords; inputting the search keywords into a search engine, and generating corresponding search results; designing a third prompt based on the search results, inputting the third prompt, the target question, the search plan, and the search results into the LLM, and generating corresponding knowledge text; designing a fourth prompt based on the knowledge text, inputting the fourth prompt, the target question, and the knowledge text into the LLM, and generating a corresponding target answer. In this way, the intelligent question-answering method of the present invention can effectively improve the accuracy, comprehensiveness, user experience, and ability to handle complex affairs during question-answering. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0052] Figure 1 A flowchart of an intelligent question-answering method provided by an embodiment of the present invention.

[0053] Figure 2 A schematic diagram of functional modules of an intelligent question-answering system provided in an embodiment of the present invention.

[0054] Figure 3 A schematic diagram of the hardware structure of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solution and effect of the present invention clearer and more specific, the present invention is further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0056] In the description of the present invention, the terms "include", "including", "have", "contain", etc. are all open terms, which mean including but not limited to. The descriptions with reference to the terms "one embodiment", "a specific embodiment", "some embodiments", "for example", etc. mean that the specific features, structures or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. The order of steps involved in each embodiment is used to schematically illustrate the implementation of the present application, and the order of steps therein is not limited and can be appropriately adjusted as needed.

[0057] Various non-limiting embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0058] At present, with the rapid development of information technology, the application of artificial intelligence in various fields has gradually deepened, and the insurance industry is no exception. At present, the existing insurance question-and-answer systems are generally divided into two categories: FAQ insurance question-and-answer and knowledge base insurance question-and-answer. Although these two systems have made certain progress in providing insurance consulting services to users, there are still some problems and limitations, such as:

[0059] 1. Inaccurate semantic understanding leads to inaccurate answers;

[0060] Semantic understanding is the key to accurately answering user questions in insurance question-answering systems. For example, existing FAQ insurance question-answering systems mostly use similarity matching methods, that is, selecting the most appropriate answer by comparing the text similarity between user questions and questions in the knowledge base. However, this method often ignores the deep semantics and implicit information behind user questions, resulting in low accuracy of answers.

[0061] 2. The quality of the knowledge base is not high, resulting in incomplete answers;

[0062] The knowledge base is the core component of the FAQ insurance question-and-answer system. Its quality and comprehensiveness directly affect the accuracy and comprehensiveness of the system's answers. The current insurance question-and-answer system knowledge base often has problems such as untimely information updates and limited coverage, which results in the system being unable to provide satisfactory answers when faced with some specific or novel questions. This not only affects the user experience, but also limits the system's service capabilities.

[0063] 3. The conversation process is not natural enough, resulting in poor user experience;

[0064] User experience is another important indicator to measure the success of insurance Q&A systems. Existing insurance Q&A systems are often too simple in dialogue process design and lack a natural and smooth interactive experience. When users communicate with the system, they may encounter comprehension barriers or poor communication, which greatly reduces user satisfaction and the practicality of the system.

[0065] 4. Insufficient ability to handle complex issues and multi-round conversations;

[0066] When faced with complex questions or scenarios that require multiple rounds of dialogue, existing FAQ question-answering systems often seem to be unable to cope. They can usually only handle simple questions and answers, and cannot conduct in-depth logical reasoning or continuous dialogue exchanges. For example, when dealing with insurance issues involving specific drug claims, the system may not be able to accurately determine the correlation between the drug and the accident, resulting in an inability to give reasonable claims recommendations.

[0067] Based on this, how to provide an intelligent question-answering method, system, computer device and non-volatile computer-readable storage medium to effectively improve the accuracy, comprehensiveness, user experience and ability to handle complex transactions during question-answering is a problem that urgently needs to be solved by technical personnel in this field.

[0068] Therefore, to solve the above problem, please refer to Figure 1 The embodiment of the present invention provides an intelligent question-answering method, wherein the method comprises the steps of:

[0069] S100, designing a first prompt based on a target question given by a target user, inputting the first prompt and the target question into a large language model LLM, and generating a corresponding search plan;

[0070] S200, designing a second prompt based on the search plan, inputting the second prompt and the search plan into the LLM, and generating corresponding search keywords;

[0071] S300, inputting the search keyword into a search engine to generate corresponding search results;

[0072] S400, designing a third prompt based on the search results, inputting the third prompt, the target question, the search plan, and the search results into the LLM, and generating a corresponding knowledge text;

[0073] S500, designing a fourth prompt based on the knowledge text, inputting the fourth prompt, the target question, and the knowledge text into the LLM, and generating a corresponding target answer.

[0074] Furthermore, the intelligent question answering method, wherein the step S100, designs a first prompt based on a target question given by a target user, inputs the first prompt and the target question into a large language model LLM, and generates a corresponding search plan, includes:

[0075] Get the target question input by the target user;

[0076] Designing a first initial prompt based on the target question given by the target user, and obtaining a corresponding first prompt after debugging the first initial prompt multiple times;

[0077] The first prompt and the target question are input into a preset large language model LLM to generate a corresponding search plan.

[0078] In the specific implementation, in this embodiment, a first prompt is designed based on a target question given by a target user, the first prompt and the target question are input into a large language model LLM, and a corresponding search plan is generated. The specific implementation process is as follows:

[0079] 1) Target user input acquisition

[0080] Design a user-friendly interface that allows users to input their questions in natural language. This interface should be able to handle different types of input, including text, voice, etc.

[0081] 2) Design the first initial prompt

[0082] Design the first initial prompt based on the target question input by the target user; this first initial prompt should be able to clearly express the target question of the target user and guide the large language model (LLM) to understand the intention and key points of the target question;

[0083] 3) First Initial Prompt Debug

[0084] Perform multiple debugging on the first initial prompt to ensure that it can effectively interact with the large language model; this may include adjusting the wording, structure, and amount of information contained in the first initial prompt to improve its accuracy and efficiency;

[0085] 4) Generate a search plan

[0086] The debugged first prompt and the target question are input into the preset large language model LLM (such as GPT-4); the large language model will generate a corresponding search plan based on the content of the first prompt and the target question; this search plan will guide how to further refine the question to obtain a more accurate answer;

[0087] The key to this process is to ensure that the design and debugging of the first prompt can accurately capture the target user's intention for the target question and effectively guide the large language model to perform information retrieval and question answering to generate the corresponding search plan;

[0088] For example:

[0089] (1) Design the first prompt: "You are a professional working in the insurance field, using LLM. Currently, you can only answer the target question through search engines and your own knowledge accumulation. Based on the target question and the means to answer it, develop a plan to answer the target question.";

[0090] (2) Input the first prompt and the target question into LLM, and use LLM to generate the corresponding search plan.

[0091] Furthermore, the intelligent question-answering method, wherein the step S200, designing a second prompt based on the search plan, inputting the second prompt and the search plan into the LLM, and generating corresponding search keywords, comprises:

[0092] Obtaining the search plan generated by the LLM;

[0093] Designing a second initial prompt based on the search plan, and obtaining a corresponding second prompt after multiple debugging of the second initial prompt;

[0094] The second prompt and the search plan are input into the LLM to generate corresponding search keywords.

[0095] In specific implementation, in this embodiment, a second prompt is designed based on the search plan, and the second prompt and the search plan are input into the LLM to generate corresponding search keywords. The specific implementation process is as follows:

[0096] 1) Get the search plan generated by LLM

[0097] After the LLM generates a corresponding search plan according to the first prompt and the content of the target question, the search plan is obtained;

[0098] 2) Design the second initial prompt

[0099] Design a second initial prompt based on the search plan. This second initial prompt should be able to further refine the question and clarify the direction and focus of the search.

[0100] 3) Second initial prompt debugging

[0101] Debug the second initial prompt multiple times to ensure that it can effectively interact with the LLM and guide the LLM to generate more accurate search keywords; the debugging process may include adjusting the wording, structure, and amount of information contained in the second initial prompt to improve its accuracy and efficiency;

[0102] 4) Generate search keywords

[0103] Input the debugged second prompt and the search plan into LLM; LLM will generate corresponding search keywords based on the second prompt and the search plan; these keywords will be used to guide more targeted searches in the search engine;

[0104] Through this process, the search plan generated by LLM can be used more effectively to generate more precise search keywords, which helps to directly locate information and reduce the interference of irrelevant search results to provide more accurate and comprehensive answers;

[0105] For example:

[0106] (1) Design the second prompt: "You are a professional working in the insurance field. Use LLM to convert the search plan into text that can be searched by search engines based on the content of the target question.";

[0107] (2) Enter the second prompt and search plan, and use LLM to generate the corresponding search keywords.

[0108] Furthermore, in the intelligent question-answering method, the step S300 of inputting the search keyword into a search engine to generate corresponding search results includes:

[0109] The search keyword is input into a preset search engine to perform a search and generate corresponding search results.

[0110] In the specific implementation, in this embodiment, the search keyword is input into the search engine to generate corresponding search results, wherein the specific implementation process is as follows:

[0111] 1) Determine the search engine

[0112] Choose a suitable search engine, which can be a general search engine such as Google, or a vertical search engine customized specifically for the insurance field;

[0113] 2) Format search keywords

[0114] Format the search keywords according to the requirements of the selected search engine; this may include adjusting the order of the keywords, adding specific search operators, or using Boolean logic to refine the search criteria;

[0115] 3) Perform a search

[0116] Entering the formatted search keyword into the search engine to perform a search operation; this will trigger the search engine to search its database or network resources to find information related to the search keyword;

[0117] 4) Get search results

[0118] The search engine will return a series of search results, which usually include web links, documents, images or other forms of related content;

[0119] 5) Evaluate and filter search results

[0120] Perform a preliminary evaluation of the search results to select the information most relevant to the target question; this may involve a quick scan of the title, summary or content of the search results;

[0121] 6) In-depth analysis

[0122] Conduct in-depth analysis of the filtered search results to determine whether they provide the answers or relevant information the target user needs; this may require reading the entire document or web page to obtain more detailed information;

[0123] 7) Results Integration

[0124] Integrate the information obtained from search results into a comprehensive answer to the target question; this may involve extracting key information from multiple sources and organizing it into a coherent, understandable whole;

[0125] Through this process, the powerful functions of search engines can be utilized to quickly and accurately find the information required by target users by searching keywords to generate corresponding search results.

[0126] Furthermore, the intelligent question-answering method, wherein the step S400, designs a third prompt based on the search results, inputs the third prompt, the target question, the search plan, and the search results into the LLM, and generates a corresponding knowledge text, includes:

[0127] Obtaining the search results generated by the search engine;

[0128] Designing a third initial prompt based on the search result, and obtaining a corresponding third prompt after debugging the third initial prompt multiple times;

[0129] The third prompt, the target question, the search plan, and the search results are input into the LLM to generate a corresponding knowledge text.

[0130] In specific implementation, in this embodiment, a third prompt is designed based on the search results, and the third prompt, the target question, the search plan, and the search results are input into the LLM to generate a corresponding knowledge text, wherein the specific implementation process is as follows:

[0131] 1) Get search results

[0132] Obtain search results for search keywords from a preset search engine; these results may include web pages, documents, articles, and other forms of information;

[0133] 2) Design the third initial prompt

[0134] Design a third initial prompt based on the search results; this third initial prompt should be able to guide LLM to analyze and integrate the search results more deeply to generate a more comprehensive knowledge text;

[0135] 3) Third initial prompt debugging

[0136] Debug the third initial prompt multiple times to ensure that it can effectively guide LLM to process search results and generate high-quality knowledge text; the debugging process may include adjusting the wording, structure and amount of information contained in the third initial prompt to improve its accuracy and efficiency;

[0137] 4) Generate knowledge text

[0138] Input the debugged third prompt, target question, search plan and search results into LLM; LLM will integrate the information based on these inputs and generate corresponding knowledge text; this knowledge text should be a detailed answer to the user's question and combine the relevant information in the search results;

[0139] Through this process, the search engine search results can be used more effectively, combined with the powerful processing capabilities of LLM, to generate accurate, comprehensive and user-friendly high-quality knowledge texts, thereby providing high-quality insurance consulting services;

[0140] For example:

[0141] (1) Design the third prompt: "You are a professional working in the insurance field. Use LLM to extract text that is highly relevant to the target question and search plan based on the search results output by the search engine.";

[0142] (2) Input the third prompt, target question, search plan, and search results, and use LLM to generate high-quality knowledge text that can answer the target question.

[0143] Furthermore, the intelligent question-answering method, wherein the step S500, designing a fourth prompt based on the knowledge text, inputting the fourth prompt, the target question, and the knowledge text into the LLM, and generating a corresponding target answer, includes:

[0144] Obtaining the knowledge text generated by the LLM;

[0145] Designing a fourth initial prompt based on the knowledge text, and obtaining a corresponding fourth prompt after multiple debugging of the fourth initial prompt;

[0146] The fourth prompt, the target question, and the knowledge text are input into the LLM to generate a corresponding target answer.

[0147] In specific implementation, in this embodiment, the fourth prompt is designed based on the knowledge text, the fourth prompt, the target question, and the knowledge text are input into the LLM, and the corresponding target answer is generated. The specific implementation process is as follows:

[0148] 1) Obtain knowledge text

[0149] Get the generated knowledge text from LLM, which is the information synthesized based on user questions, search plans and search results;

[0150] 2) Analyze knowledge text

[0151] Analyze the knowledge text to ensure that it contains all the information relevant to the target problem and that the information is accurate and reliable;

[0152] 3) Design the fourth initial prompt

[0153] Design the fourth initial prompt based on the knowledge text. This fourth initial prompt should be able to guide the LLM to refine and summarize the information in the knowledge text to generate a concise, accurate and direct answer;

[0154] 4) Fourth initial prompt debugging

[0155] Debug the fourth initial prompt multiple times to ensure that it can effectively guide the LLM to generate the target answer; the debugging process may include adjusting the wording, structure and amount of information contained in the fourth initial prompt to improve its accuracy and efficiency;

[0156] 5) Generate target answer

[0157] Input the debugged fourth prompt, target question and knowledge text into LLM. LLM will extract and summarize the information in the knowledge text based on these inputs and generate the final target answer.

[0158] 6) Evaluate the target answer

[0159] Evaluate the target answers generated by the LLM to ensure their accuracy, relevance, and completeness; the evaluation may include checking whether the target answers directly answer the target questions of the target users and whether they contain all necessary information;

[0160] Through this process, we can leverage the power of LLM, combined with the knowledge text of the search engine, to generate high-quality, accurate and user-friendly targeted answers, thereby providing excellent insurance consulting services;

[0161] For example:

[0162] (1) Design the fourth prompt: "Use LLM to answer the target question based on the knowledge text and your own knowledge and output the target answer."

[0163] (2) Input the fourth prompt, the target question, and the knowledge text, and use LLM to generate the target answer corresponding to the target question.

[0164] Furthermore, the intelligent question-answering method, wherein the fourth prompt is designed based on the knowledge text, the fourth prompt, the target question, and the knowledge text are input into the LLM, and after the corresponding target answer is generated, further comprises:

[0165] The target answer is returned to the target user.

[0166] In the specific implementation, in this embodiment, the target answer is returned to the target user, wherein the specific implementation process is as follows:

[0167] 1) Answer formatting

[0168] Ensure that the target answer is formatted appropriately for the target audience to read and understand. This may include adjusting the formatting of the text, using clear language, and a logical structure.

[0169] 2) Target answer review

[0170] Before sending the target answer to the target user, conduct a final review to ensure that the target answer does not contain any missing, wrong or inaccurate information;

[0171] 3) Personalization

[0172] If possible, personalize the target answer based on the target user's preferences or previous interaction history to improve the target user's satisfaction;

[0173] 4) User interface presentation

[0174] Present the target answer to the target user through the user interface (UI), ensuring that the target answer is presented clearly and easily readable in the UI and is directly related to the target user’s target question;

[0175] 5) Interactive Feedback

[0176] Provide a way for target users to easily provide feedback, which can be a simple "satisfied" or "unsatisfied" button, or a more detailed feedback form;

[0177] 6) Feedback Collection

[0178] Collecting feedback from target users on target answers is crucial for subsequent improvements and can help identify areas for improvement.

[0179] 7) Answer Record

[0180] Record the target questions and answers of target users for subsequent analysis and optimization;

[0181] 8) Follow-up

[0182] If the target user's target question requires further follow-up or clarification, ensure that you can provide additional support or guide the target user to obtain more information;

[0183] 9) Optimization and learning

[0184] Leverage user feedback and answer records to optimize and learn, in order to improve the accuracy and efficiency of future Q&A;

[0185] Through this process, not only can accurate answers be provided to target users, but valuable user feedback can also be collected for continuous improvement and optimization; this closed-loop question-and-answer process helps to improve user satisfaction and ensure continuous improvement over time.

[0186] It can be seen from the above method embodiments that the intelligent question-answering method provided by the present invention at least includes: designing a first prompt based on the target question given by the target user, inputting the first prompt and the target question into the large language model LLM, and generating a corresponding search plan; designing a second prompt based on the search plan, inputting the second prompt and the search plan into the LLM, and generating corresponding search keywords; inputting the search keywords into the search engine, and generating corresponding search results; designing a third prompt based on the search results, inputting the third prompt, the target question, the search plan, and the search results into the LLM, and generating corresponding knowledge text; designing a fourth prompt based on the knowledge text, inputting the fourth prompt, the target question, and the knowledge text into the LLM, and generating a corresponding target answer; and so on. In this way, the intelligent question-answering method of the present invention can effectively improve the accuracy, comprehensiveness, user experience, and ability to handle complex affairs during question-answering.

[0187] Specifically, the intelligent question-answering method provided by the embodiment of the present invention has at least the following beneficial technical effects:

[0188] 1. The knowledge learned is more comprehensive: After fine-tuning with insurance knowledge, LLM not only possesses professional and rich insurance knowledge, but also possesses a vast amount of human knowledge, with a wider range of knowledge and a greatly increased breadth of answering questions;

[0189] 2. A stronger ability to understand natural language helps to fully extract the potential semantic information of user questions, thereby improving the accuracy of answering questions;

[0190] 3. According to the characteristics of the insurance question-answering system, four prompts are designed to better adapt to the insurance field, LLM architecture and algorithm details, guide LLM to answer questions in a human way of thinking, fully understand the semantic information of user questions, and finally generate more accurate output;

[0191] 4. It is proposed to combine the search engine with the insurance question-and-answer system. Due to the knowledge limitations of the model itself and the limitations of the knowledge base, some tricky questions cannot be processed and answered. Therefore, when constructing the question-and-answer system, the present invention introduces a search engine to search for knowledge online, which can broaden the scope of answering questions and improve the accuracy of answering questions.

[0192] It should be understood that, although the present application provides method operation steps as described in the embodiments or flowcharts, more or less operation steps may be included based on conventional or non-creative labor, and these operation steps are not necessarily performed in sequence according to the embodiment or flowchart. The order of steps listed in the embodiment or flowchart is only one way of executing the order of many steps, and does not represent the only execution order. It should be noted that there is not necessarily a certain order between the above steps. A person of ordinary skill in the art can understand from the description of the embodiment of the present invention that in different embodiments, the above steps may have different execution orders, that is, they may be executed in parallel, or they may be executed in exchange, etc. Moreover, at least a part of the steps in the embodiment or flowchart may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but may be executed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but may be executed in turn, alternately or synchronously with other steps or at least a part of the sub-steps or stages of other steps.

[0193] Based on the above method embodiment, please refer to Figure 2 Another embodiment of the present invention further provides an intelligent question-answering system, wherein the system comprises:

[0194] A first generating module 11 is used to design a first prompt based on a target question given by a target user, input the first prompt and the target question into a large language model LLM, and generate a corresponding search plan;

[0195] A second generating module 12 is used to design a second prompt based on the search plan, input the second prompt and the search plan into the LLM, and generate corresponding search keywords;

[0196] Search module 13, used to input the search keyword into a search engine to generate corresponding search results;

[0197] A third generating module 14 is used to design a third prompt based on the search results, input the third prompt, the target question, the search plan, and the search results into the LLM, and generate a corresponding knowledge text;

[0198] The fourth generating module 15 is used to design a fourth prompt based on the knowledge text, input the fourth prompt, the target question, and the knowledge text into the LLM, and generate a corresponding target answer.

[0199] The specific implementation method is shown in the above method embodiment, which will not be described again here.

[0200] It can be seen from the above system embodiments that the intelligent question-answering system provided by the present invention at least includes: designing a first prompt based on the target question given by the target user, inputting the first prompt and the target question into the large language model LLM, and generating a corresponding search plan; designing a second prompt based on the search plan, inputting the second prompt and the search plan into the LLM, and generating corresponding search keywords; inputting the search keywords into the search engine, and generating corresponding search results; designing a third prompt based on the search results, inputting the third prompt, the target question, the search plan, and the search results into the LLM, and generating corresponding knowledge text; designing a fourth prompt based on the knowledge text, inputting the fourth prompt, the target question, and the knowledge text into the LLM, and generating a corresponding target answer; and so on. In this way, the intelligent question-answering system of the present invention can effectively improve the accuracy, comprehensiveness, user experience, and ability to handle complex affairs when answering questions.

[0201] Specifically, the intelligent question-answering system provided by the embodiment of the present invention has at least the following beneficial technical effects:

[0202] 1. The knowledge learned is more comprehensive: After fine-tuning with insurance knowledge, LLM not only possesses professional and rich insurance knowledge, but also possesses a vast amount of human knowledge, with a wider range of knowledge and a greatly increased breadth of answering questions;

[0203] 2. A stronger ability to understand natural language helps to fully extract the potential semantic information of user questions, thereby improving the accuracy of answering questions;

[0204] 3. According to the characteristics of the insurance question-answering system, four prompts are designed to better adapt to the insurance field, LLM architecture and algorithm details, guide LLM to answer questions in a human way of thinking, fully understand the semantic information of user questions, and finally generate more accurate output;

[0205] 4. It is proposed to combine the search engine with the insurance question-and-answer system. Due to the knowledge limitations of the model itself and the limitations of the knowledge base, some tricky questions cannot be processed and answered. Therefore, when constructing the question-and-answer system, the present invention introduces a search engine to search for knowledge online, which can broaden the scope of answering questions and improve the accuracy of answering questions.

[0206] Based on the above method embodiment, please refer to Figure 3 Another embodiment of the present invention further provides a computer device, wherein the computer device 10 includes:

[0207] memory 120 and one or more processors 110, Figure 3A processor 110 is used as an example for the description. The processor 110 and the memory 120 may be connected via a communication bus or other means. Figure 3 The example of connecting via a communication bus is taken as an example.

[0208] The processor 110 is used to complete various control logics of the computer device 10. It can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), a single-chip microcomputer, an ARM (Acorn RISC Machine) or other programmable logic device, a discrete gate or transistor logic, a discrete hardware component, or any combination of these components. In addition, the processor 110 can also be any traditional processor, microprocessor or state machine. The processor 110 can also be implemented as a combination of computing devices, for example, a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors combined with a DSP, or any other such configuration.

[0209] The memory 120 is a non-volatile computer-readable storage medium that can be used to store non-volatile software programs, non-volatile computer executable programs and modules, such as the computer program corresponding to the intelligent question-answering method in the embodiment of the present invention. The processor 110 executes various functional applications and data processing of the computer device 10 by running the non-volatile software programs, instructions and units stored in the memory 120, that is, implementing the intelligent question-answering method in the above method embodiment.

[0210] The memory 120 may include a program storage area and a data storage area, wherein the program storage area may store an application required for operating the device or at least one function; the data storage area may store data created according to the use of the computer device 10, etc. In addition, the memory 120 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 120 may optionally include a memory remotely arranged relative to the processor 110, and these remote memories may be connected to the computer device 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0211] One or more units are stored in the memory 120, and when executed by one or more processors 110, the intelligent question-answering method in any of the above method embodiments can be implemented, for example, the intelligent question-answering method described above can be implemented. Figure 1 The method comprises steps S100 to S500.

[0212] Those skilled in the art will understand that Figure 3The hardware structure diagram shown in the figure is only a schematic diagram of a partial structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied. The specific computer device may include more components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0213] Based on the above method embodiments, another embodiment of the present invention further provides a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by at least one processor, the intelligent question-answering method in any of the above method embodiments can be implemented, for example, the above-described Figure 1 The method comprises steps S100 to S500.

[0214] As an example, the non-volatile storage medium can include a read-only memory (ROM), a programmable ROM (PROM), an electrically programmable ROM (EPROM), an electrically erasable ROM (EEPROM), or a flash memory. Volatile memory can include a random access memory (RAM) as an external cache memory. By way of illustration and not limitation, RAM can be obtained in many forms such as synchronous RAM (SRAM), dynamic RAM, (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), Synch link DRAM (SLDRAM), and direct Rambus RAM (DRRAM). The disclosed memory components or memories of the operating environment described herein are intended to include one or more of these and / or any other suitable types of memory.

[0215] Based on the above method embodiments, another embodiment of the present invention further provides a computer program product, the computer program product includes a computer program stored on a non-volatile computer-readable storage medium, the computer program includes program instructions, when the program instructions are executed by a processor, the intelligent question answering method in any of the above method embodiments can be implemented, for example, the above described Figure 1 The method comprises steps S100 to S500.

[0216] The embodiments described above are merely illustrative, in which the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0217] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a general hardware platform, and of course can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the relevant technology can be embodied in the form of a software product, and the computer software product can exist in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of each embodiment or some parts of the embodiment.

[0218] Conditional language such as "can," "can," "might," or "may," unless specifically stated otherwise or otherwise understood within the context as used, is generally intended to convey, among other things, that a particular embodiment can include (while other embodiments do not) a particular feature, element, and / or operation. Thus, such conditional language is also generally intended to imply that a feature, element, and / or operation is required for one or more embodiments anyway or that one or more embodiments must include logic for determining, with or without input or prompting, whether such features, elements, and / or operations are included or will be performed in any particular embodiment.

[0219] What has been described herein, in this specification and in the accompanying drawings includes examples that can provide an intelligent question-answering method, system, computer device, and non-volatile computer-readable storage medium. Of course, it is not possible to describe every conceivable combination of elements and / or methods for the purpose of describing the various features of the present disclosure, but it can be recognized that many other combinations and permutations of the disclosed features are possible. Therefore, it is obvious that various modifications can be made to the present disclosure without departing from the scope or spirit of the present disclosure, but all these various modifications should belong to the scope of protection of the claims attached to the present invention. In addition, or in an alternative, other embodiments of the present disclosure may be obvious from consideration of the present specification and the accompanying drawings and from the practice of the present disclosure as presented herein. It is intended that the examples proposed in the present specification and the accompanying drawings are considered to be illustrative and non-restrictive in all respects. Although specific terms are used herein, they are used in a general and descriptive sense and are not used for limiting purposes.

Claims

1. An intelligent question-answering method, characterized in that: include: Designing a first prompt based on a target question given by a target user, inputting the first prompt and the target question into a large language model LLM, and generating a corresponding search plan; Design a second prompt based on the search plan, input the second prompt and the search plan into the LLM, and generate corresponding search keywords; Input the search keyword into a search engine to generate corresponding search results; Designing a third prompt based on the search results, inputting the third prompt, the target question, the search plan, and the search results into the LLM to generate a corresponding knowledge text; A fourth prompt is designed based on the knowledge text, and the fourth prompt, the target question, and the knowledge text are input into the LLM to generate a corresponding target answer.

2. The intelligent question-answering method according to claim 1, characterized in that: The step of designing a first prompt based on a target question given by a target user, inputting the first prompt and the target question into a large language model (LLM), and generating a corresponding search plan includes: Get the target question input by the target user; Designing a first initial prompt based on the target question given by the target user, and obtaining a corresponding first prompt after debugging the first initial prompt multiple times; The first prompt and the target question are input into a preset large language model LLM to generate a corresponding search plan.

3. The intelligent question-answering method according to claim 2, characterized in that: The step of designing a second prompt based on the search plan, inputting the second prompt and the search plan into the LLM, and generating corresponding search keywords includes: Obtaining the search plan generated by the LLM; Designing a second initial prompt based on the search plan, and obtaining a corresponding second prompt after multiple debugging of the second initial prompt; The second prompt and the search plan are input into the LLM to generate corresponding search keywords.

4. The intelligent question-answering method according to claim 3, characterized in that: The step of inputting the search keyword into a search engine to generate corresponding search results includes: The search keyword is input into a preset search engine to perform a search and generate corresponding search results.

5. The intelligent question-answering method according to claim 4, characterized in that: The step of designing a third prompt based on the search results, inputting the third prompt, the target question, the search plan, and the search results into the LLM, and generating a corresponding knowledge text includes: Obtaining the search results generated by the search engine; Designing a third initial prompt based on the search result, and obtaining a corresponding third prompt after debugging the third initial prompt multiple times; The third prompt, the target question, the search plan, and the search results are input into the LLM to generate a corresponding knowledge text.

6. The intelligent question-answering method according to claim 5, characterized in that: The designing of the fourth prompt based on the knowledge text, inputting the fourth prompt, the target question, and the knowledge text into the LLM, and generating a corresponding target answer includes: Obtaining the knowledge text generated by the LLM; Designing a fourth initial prompt based on the knowledge text, and obtaining a corresponding fourth prompt after multiple debugging of the fourth initial prompt; The fourth prompt, the target question, and the knowledge text are input into the LLM to generate a corresponding target answer.

7. The intelligent question-answering method according to any one of claims 1 to 6, characterized in that: After designing a fourth prompt based on the knowledge text, inputting the fourth prompt, the target question, and the knowledge text into the LLM, and generating a corresponding target answer, the method further includes: The target answer is returned to the target user.

8. An intelligent question-answering system, characterized in that: include: A first generating module is used to design a first prompt based on a target question given by a target user, input the first prompt and the target question into a large language model LLM, and generate a corresponding search plan; A second generating module, configured to design a second prompt based on the search plan, input the second prompt and the search plan into the LLM, and generate corresponding search keywords; A search module, used to input the search keyword into a search engine to generate corresponding search results; A third generation module is used to design a third prompt based on the search results, input the third prompt, the target question, the search plan, and the search results into the LLM, and generate a corresponding knowledge text; The fourth generation module is used to design a fourth prompt based on the knowledge text, input the fourth prompt, the target question, and the knowledge text into the LLM, and generate a corresponding target answer.

9. A computer device, characterized in that: The computer device comprises at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor. When the computer program is executed by the at least one processor, the intelligent question and answer method as described in any one of claims 1 to 7 can be implemented.

10. A non-volatile computer-readable storage medium, characterized in that: The non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by at least one processor, the intelligent question-answering method according to any one of claims 1 to 7 can be implemented.

Citation Information

Patent Citations

  • Knowledge question and answer method, device and equipment and storage medium

    CN116561278A

  • Method and system for optimizing cue word, electronic equipment and storage medium

    CN118428492A

Cited By

  • Digital person business card interaction method, electronic equipment and storage medium

    CN119398728A

  • Searching method and device, program product and storage medium

    CN119961420A