An intelligent question answering system and method based on large language model

Abstract

The present application discloses an intelligent question-answering system and method based on a large language model, and relates to the field of intelligent question-answering systems. The intelligent question-answering system includes: a user question acquisition module, a context processing module, a question classification module, a knowledge base management module, a question-answering processing module, and a reply module. The user question acquisition module acquires the user's current question, the context processing module adopts a large language model, and generates a search statement according to the user's historical question information and the user's current question. The question classification module classifies the user's question according to the search statement, the knowledge base management module stores technical solutions, the question-answering processing module retrieves the technical solutions stored in the knowledge base management module based on the question classification results, and the reply module adopts a large language model to generate reply content based on the retrieved technical solutions. The present application can achieve efficient and accurate answers to customers' technical inquiries, and can automatically generate personalized solutions according to the specific needs of customers.

Description

Technical Field

[0001] The present application relates to the field of intelligent customer service systems, and in particular to an intelligent question-answering system and method based on a large language model. Background Art

[0002] With the rapid development of the global economy, product counterfeiting and imitation are becoming more and more serious. Anti-counterfeiting technology has become an urgent issue facing all industries and is crucial to protecting brand reputation, market order and consumer rights. However, the current customer service system is unable to cope with complex anti-counterfeiting technical issues. Traditional customer service often relies on manual question answering, which is not only limited in professionalism, but also slow in response speed. At the same time, existing anti-counterfeiting solutions are mostly based on manual inspection, resulting in low work efficiency and high error rate, which cannot meet customers' requirements for timeliness and accuracy. Summary of the invention

[0003] The purpose of this application is to provide an intelligent question-answering system and method based on a large language model, which can efficiently and accurately answer customers' technical inquiries, and automatically generate personalized solutions based on the specific needs of customers.

[0004] To achieve the above objectives, this application provides the following solutions:

[0005] In a first aspect, the present application provides an intelligent question-answering system based on a large language model, the intelligent question-answering system based on a large language model comprising:

[0006] User question acquisition module, used to obtain the user's current question;

[0007] A context processing module, connected to the user question acquisition module, for using a large language model to generate a search statement based on the user's historical question information and the user's current question;

[0008] A question classification module, connected to the context processing module and the user question acquisition module, is used to classify the user's current question according to the search statement to obtain a variety of question classification results; the question classification results include: technical consulting questions, solution consulting questions, and questions other than technical consulting questions and solution consulting questions;

[0009] A knowledge base management module is used to store technical solutions; the technical solutions include: technical knowledge content and application cases of the technical knowledge content;

[0010] A question-answer processing module, connected to the user question acquisition module, the question classification module and the knowledge base management module, for retrieving the technical solution stored in the knowledge base management module based on the question classification result;

[0011] The reply module is connected to the question-answer processing module and is used to adopt the large language model to generate reply content based on the retrieved technical solution.

[0012] Optionally, the intelligent question answering system based on the large language model further includes:

[0013] The visualization module is connected to the reply module and is used to visualize and display the reply content.

[0014] Optionally, the context processing module includes:

[0015] User history question query unit, used to query user history question information;

[0016] The search statement generation unit is connected to the user question acquisition module and the user history question query unit, and is used to use the large language model to generate a search statement carrying context information based on the user history question information and the user's current question.

[0017] Optionally, the knowledge base management module includes:

[0018] A data storage unit, used to store a technical solution; the technical solution includes: technical knowledge content and application cases of the technical knowledge content;

[0019] An updating and maintaining unit is connected to the data storage unit and is used to update and maintain the technical knowledge content and application cases of the technical knowledge content stored in the data storage unit.

[0020] Optionally, the question and answer processing module includes:

[0021] A technical consultation question processing unit, connected to the user question acquisition module, the question classification module and the knowledge base management module, for retrieving the technical knowledge content stored in the knowledge base management module according to the technical consultation question;

[0022] A solution consultation question processing unit, connected to the user question acquisition module, the question classification module and the knowledge base management module, for retrieving technical knowledge content and application cases of technical knowledge content stored in the knowledge base management module according to the solution consultation question;

[0023] The question processing unit is connected with the user question acquisition module, the question classification module and the knowledge base management module, and is used to process questions other than technical consultation questions and solution consultation questions.

[0024] Optionally, the technical consultation problem processing unit includes:

[0025] The keyword generation subunit is connected to the user question acquisition module and is used to generate search keywords based on the user's historical question information and the user's current question using a large language model;

[0026] The technical search result generating subunit is connected with the question classification module, the keyword generating subunit and the knowledge base management module, and is used to retrieve the technical knowledge content stored in the knowledge base management module based on the search keyword.

[0027] Optionally, the solution consultation problem processing unit includes:

[0028] An information collection subunit, used to collect user information;

[0029] A judgment subunit, connected to the information collection subunit, for judging the integrity of the user information;

[0030] The solution retrieval result generation subunit is connected to the user question acquisition module, the question classification module, the information collection subunit, the knowledge base management module and the judgment subunit, and is used to use a large language model to retrieve the technical knowledge content and technical knowledge content application cases stored in the knowledge base management module based on the user information and the solution consultation question.

[0031] Optionally, the problem processing unit includes:

[0032] The language guidance subunit is connected to the user question acquisition module and the question classification module, and is used to use a large language model to generate guidance information based on questions other than technical consultation questions and solution consultation questions.

[0033] Optionally, the intelligent question answering system based on the large language model further includes:

[0034] The feedback module is connected to the reply module and is used to obtain user usage information and provide feedback on satisfaction with the reply content based on the user usage information.

[0035] In a second aspect, the present application provides an intelligent question answering method based on a large language model, comprising:

[0036] Get the user's historical question information and the user's current question;

[0037] Using a large language model, a search statement is generated according to the user's historical question information and the user's current question;

[0038] Classify the user's current question according to the search statement to obtain multiple question classification results; the question classification results include: technical consulting questions, solution consulting questions, and questions other than technical consulting questions and solution consulting questions;

[0039] Retrieving the technical solutions stored in the knowledge base based on the problem classification result;

[0040] The large language model is used to generate reply content based on the retrieved technical solution.

[0041] According to the specific embodiments provided in this application, this application has the following technical effects:

[0042] The present application provides an intelligent question-answering system and method based on a large language model, which generates a search statement based on the user's historical question information and the user's current question, and then classifies the user's current question according to the search statement, and obtains a variety of question classification results, so as to facilitate the retrieval of stored technical solutions from the knowledge base, improve the accuracy of question answers, and realize personalized recommendations; and, by adopting a large language model, the reply content is generated based on the retrieved technical solution, which can efficiently and accurately answer customers' technical consultations, and can automatically generate personalized solutions according to the customer's specific question needs, thereby solving the problems of traditional customer service relying on manual problem answering, lack of professionalism, low work efficiency and high error rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0044] Figure 1 This is a structural diagram of an intelligent question-answering system based on a large language model in one embodiment of the present application;

[0045] Figure 2 A schematic diagram of a common user problem handling process provided in an embodiment of the present application;

[0046] Figure 3 A schematic diagram of a user question processing flow of an intelligent question answering system based on a large language model provided in an embodiment of the present application;

[0047] Figure 4 A schematic diagram of a conventional user question classification process provided in an embodiment of the present application;

[0048] Figure 5 A schematic diagram of a user question classification process of an intelligent question answering system based on a large language model provided in an embodiment of the present application;

[0049] Figure 6 A schematic diagram of a knowledge base establishment process provided in an embodiment of the present application;

[0050] Figure 7 A schematic diagram of a knowledge base search process provided in an embodiment of the present application;

[0051] Figure 8 A schematic diagram of the anti-counterfeiting technology consultation process provided in one embodiment of the present application;

[0052] Fig. 9 A schematic diagram of the anti-counterfeiting solution consultation process provided in one embodiment of the present application;

[0053] Fig.10 A schematic diagram of a process for processing other types of problems provided in an embodiment of the present application;

[0054] Fig.11 A schematic diagram of the result generation and feedback process provided in an embodiment of the present application;

[0055] Fig.12 A schematic flowchart of the steps of applying an intelligent question-answering method based on a large language model in the field of anti-counterfeiting provided in one embodiment of the present application. DETAILED DESCRIPTION

[0056] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0057] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0058] In an exemplary embodiment, the present application provides an intelligent question-answering system based on a large language model, such as Figure 1 As shown, the intelligent question-answering system based on the large language model includes: a user question acquisition module 100, a context processing module 101, a question classification module 102, a knowledge base management module 103, a question-answering processing module 104 and a reply module 105.

[0059] The user question acquisition module 100 is used to acquire the user's current question.

[0060] The context processing module 101 is used to use a large language model to generate a search statement based on the user's historical question information and the user's current question.

[0061] The question classification module 102 is connected to the context processing module 101, and is mainly used to classify the user's current question according to the search statement to obtain a variety of question classification results; the question classification results include: technical consulting questions, solution consulting questions, and questions other than technical consulting questions and solution consulting questions.

[0062] The knowledge base management module 103 is used to store technical solutions; the technical solutions include: technical knowledge content and application cases of the technical knowledge content.

[0063] The question and answer processing module 104 is connected to the question classification module 102 and the knowledge base management module 103, and is used to retrieve the technical solutions stored in the knowledge base management module based on the question classification results.

[0064] The reply module 105 is connected to the question-answer processing module 104 and is used to generate reply content based on the retrieved technical solutions by using a large language model.

[0065] In order to more clearly discuss the various modules of the intelligent question-answering system based on a large language model in the present application, the following embodiments are discussed using the application of the intelligent question-answering system in the present application in the field of anti-counterfeiting as an example.

[0066] In another exemplary embodiment of the present application, in order to ensure that necessary contextual information is retained during question analysis and knowledge base retrieval so that subsequent processing is more accurate, the present application automatically performs contextual processing on user questions through a large language model.

[0067] The following is an example of a user asking a question to illustrate the implementation process of the context processing module provided by this application:

[0068] The first question the user asks is: "How is the micro-dot code technology implemented?"; and the second question the user asks is: "What are the advantages of this technology?".

[0069] Conventional treatment methods such as Figure 2 As shown in the figure, when the user inquires for the first time, the knowledge base of the implementation method is searched according to the micro-code technology; when the user inquires for the second time, the query results include the advantages of various technologies, and even the micro-code technology cannot be found due to the limit of the number of search items. It can be seen that this method of directly searching the knowledge base based on the user's current question will lose context information, resulting in deviation or even error in the search data.

[0070] In this application, when the user makes the first inquiry, the implementation method of the micro-dot code technology can be found by querying the knowledge base. When the user makes the second inquiry, the content of the first inquiry is first queried before searching, and then a query statement is generated based on the content of the second inquiry: "The advantages of micro-dot code technology", and then the knowledge base content is accurately retrieved based on this query statement. The specific processing flow is as follows: Figure 3 As shown in the figure, it can be seen that the system automatically processes the context of the user's question through the large language model, ensuring that the necessary context information is retained during question analysis and knowledge base retrieval. This context-aware mechanism can improve the accuracy of question classification and the relevance of subsequent answers.

[0071] In actual applications, based on the user's current question, the user's historical questions are first queried. The historical questions default to the top 50 current questions raised by the user. The specific number of questions is configurable. In this application, it is configured to 10 questions. The large language model is used to process them into search statements with context, and then the processed search statements are used to search the knowledge base to make the search results more accurate.

[0072] In another exemplary embodiment of the present application, in order to improve the accuracy and efficiency of question classification, the system provided by the present application uses a customized large model assistant to automatically classify questions raised by users.

[0073] The following is a comparative discussion of the conventional implementation method of problem classification and the implementation method provided by this application.

[0074] The conventional implementation of problem classification is as follows Figure 4 As shown in the figure, users need to select chat intent before chatting, such as anti-counterfeiting solution consultation, anti-counterfeiting technology consultation, or consultation on other issues, and then proceed to different processing flows according to the user's choice. It can be seen that when users switch intents, they need to switch manually, and the chat continuity experience is relatively poor.

[0075] The problem classification implementation method of this application is as follows Figure 5 As shown, through user chatting, the user's intention is automatically identified with the help of a large language model according to the user's chat context, and the corresponding processing flow is entered according to the user's intention. The different processes can be switched imperceptibly, giving the user a better chat experience. Furthermore, the classification method of the present application is not only based on the text content of the question, but also takes into account the user's historical interaction data, thereby improving the accuracy and efficiency of the classification.

[0076] In another exemplary embodiment of the present application, in order to ensure the accuracy of all technical details, the system in the present application is implemented through the establishment, updating and maintenance of a knowledge base.

[0077] In practical applications, the knowledge base includes the following contents: detailed description and classification of anti-counterfeiting technology and actual cases of anti-counterfeiting technology application. In practical applications, the knowledge base is usually used to establish, update, maintain and retrieve, including:

[0078] like Figure 6 The figure shows the flow chart of knowledge base establishment, where ETL in the knowledge base refers to the process of extraction, transformation, and loading. Extraction is to obtain raw data or text content from internal company documents, reports, and manual uploads by users. The transformation process is to clean and convert these raw contents into knowledge structures that can be recognized and organized by the knowledge base, involving text analysis, entity recognition, relationship extraction, etc. The documents are processed into small fragments of the knowledge base, and the documents can be segmented according to the selection (including characters, line breaks, and segmentation), and the document format is parsed (for example, the text segmenter (LangChain.text_splitter) tool in the programming framework of the large language model is used for lightweight markup language (markdown) format documents), and the titles and paragraph contents of each paragraph are extracted, and titles of various levels are added to each fragment to associate the relationship between the fragments and increase the relevance between the fragments. Loading is to store the processed knowledge in the knowledge base system for query, classification, search and other operations. Specifically, the processed fragmented knowledge fragments are stored in the local database, and the fragments are vectorized and stored in the Tencent vector database through the vectorization (Embedding) large model.

[0079] like Figure 7 The flowchart of knowledge base retrieval is shown in the figure. The specific process of knowledge base retrieval is as follows: First, the content is vectorized through the Embedding model according to the retrieval content. Secondly, the most relevant data is retrieved from the Tencent vector database through the Embedding content, and the Top X (customizable) data is retrieved, and the relevant knowledge fragments are extracted from the found content and returned.

[0080] In practical applications, the content of the knowledge base needs to be updated regularly to ensure that it keeps pace with the latest developments in the anti-counterfeiting industry. The updated knowledge base content should be further used as the core resource for the system to analyze problems and generate solutions, so as to provide a comprehensive data foundation for the intelligent question-answering system based on the large language model.

[0081] In a dynamically updated knowledge base module, the system ensures the accuracy of all technical details by summarizing and maintaining the company's anti-counterfeiting technology knowledge base, and can dynamically update the knowledge base according to the latest industry progress. The update of the knowledge base in this application adopts a combination of manual update and scheduled update, where manual update is to manually upload and modify the knowledge base content; scheduled update is to pull relevant content from the company's document library and maintain it in the knowledge base on a daily basis.

[0082] In another exemplary embodiment, Figure 8 As shown in the figure, the specific process of anti-counterfeiting technical consultation and processing is provided:

[0083] When users consult about anti-counterfeiting technology issues, the system will search for keywords based on the knowledge base generated by the big model perception system. The search keywords are generated by combining the user's historical questions and current questions. For example, the user asked about the advantages of the four-quadrant technology, and then asked what are the advantages and disadvantages? Then the question should be "the advantages and disadvantages of the four-quadrant technology", and this question will be used as the search keyword; the technical information related to the question will be retrieved from the knowledge base, and combined with the contextual content entered by the user, the technical consultant big model assistant will perform semantic integration processing based on the retrieved content, and generate high-quality anti-counterfeiting technology questions that match the user's needs and return them directly to the user.

[0084] Furthermore, if the user is not satisfied with the response provided by the big model assistant, this application can provide the contact information and company address of the company's business personnel, allowing the user to consult manually.

[0085] In another exemplary embodiment, Fig. 9 As shown in the figure, the specific process of anti-counterfeiting solution consultation and processing is provided:

[0086] When a user consults about anti-counterfeiting solutions, the system will first collect key information about the customer through the customer service big model assistant, including product characteristics, industry information, and demand quantity. If the information provided by the user is incomplete, the system will guide the user to supplement the key information to ensure the accuracy of the subsequent solution generation. When the information is complete, the system will retrieve the technical content in the anti-counterfeiting solution knowledge base, combine it with the product manager big model assistant, generate a personalized anti-counterfeiting solution, and provide timely feedback to the user.

[0087] In practical applications, the processing of anti-counterfeiting solution consultation can be achieved through the following steps:

[0088] The first step is user consultation plan.

[0089] The second step is to use the customer service big model to determine whether the user provides key information (product features, industry information) and collect it.

[0090] The third step is to determine whether the user has provided complete key information. The judgment is usually based on semantic analysis and knowledge reasoning, and relies on specific model training and classification rules to ensure that the user's questions or user statements can be effectively extracted and identified; if the user does not provide complete information, the system will use the customer service big model to provide language guidance content to allow the user to provide information. For example, when the user is required to provide detailed "anti-counterfeiting features" and "industry information", the big model assistant provides guidance content: Hello, in order to help you design a suitable anti-counterfeiting solution, I need to know some basic information. Please tell me your industry or what features you hope the anti-counterfeiting label has. This information will help us better understand your needs and provide corresponding anti-counterfeiting solutions.

[0091] The fourth step is to search the company's case knowledge base through key information after the customer provides complete information.

[0092] The fifth step is to generate recommended technical points through the technical expert model combined with the key information provided by the customer and company cases.

[0093] Step 6: Search the company's technical knowledge base through technical points.

[0094] The seventh step is to generate an anti-counterfeiting solution based on requirements by combining the product manager's big model with the technical knowledge base and key information provided by the customer.

[0095] In another exemplary embodiment, in order to handle user inquiries that are not related to anti-counterfeiting technology and anti-counterfeiting solutions, the present application provides a specific process for handling such inquiries, such as Fig.10 shown.

[0096] For user inquiries that are not related to anti-counterfeiting technology and anti-counterfeiting solutions, the system will analyze the user's questions, guide the user to inquire about professional issues, and provide the company's business contact information. For example, when a user asks about today's weather, the assistant replies: "Hello! I don't know much about today's weather conditions, because my main responsibility is to provide support for anti-counterfeiting technology and related services. You can check the local weather forecast to get the most accurate information. If you have any questions about anti-counterfeiting technology or other related issues, I am happy to answer them for you. Is there anything I can help you with?" 。

[0097] In another exemplary embodiment, in order to timely understand the user's satisfaction with the reply content and improve the user's participation, the present application provides a user feedback module. The feedback process is as follows: Fig.11 shown.

[0098] The technical customer service model will conduct a comprehensive analysis based on the content extracted from the knowledge base and user questions to generate detailed and targeted technical answers or solutions. After the system generates a technical answer or solution, the user can provide feedback. For example, the user can click the evaluation button on the reply interface: agree or disagree. The administrator can adjust and maintain the knowledge base content based on the feedback results.

[0099] In another exemplary embodiment, the present application also provides an intelligent question answering method based on a large language model, comprising:

[0100] Get the user's historical question information and the user's current question.

[0101] A large language model is used to generate a search statement based on the user's historical question information and the user's current question.

[0102] The user's current question is classified according to the search statement to obtain a variety of question classification results; the question classification results include: technical consulting questions, solution consulting questions and questions other than technical consulting questions and solution consulting questions.

[0103] The technical solutions stored in the knowledge base are retrieved based on the problem classification result.

[0104] The large language model is used to generate reply content based on the retrieved technical solution.

[0105] In the field of anti-counterfeiting, the specific implementation flow chart of an intelligent question-answering method based on a large language model provided by this application is as follows: Fig.12As shown. First, a large language model is used to generate a search statement based on the user's questions about anti-counterfeiting and the user's historical questions about anti-counterfeiting. The classification large model assistant classifies the user's questions about anti-counterfeiting according to the search statement, and can classify anti-counterfeiting technology consultation, anti-counterfeiting solution consultation and other problem consultation. Secondly, the technical solutions stored in the knowledge base are retrieved according to different problem classification results. Specifically, when the user's question belongs to the anti-counterfeiting technology consultation category, a large language model is used to process the search keywords carrying context according to the user's questions about anti-counterfeiting and the user's historical questions about anti-counterfeiting, and the anti-counterfeiting knowledge base is searched according to the search keywords. The technical consultant large model provides answers based on the retrieved content; when the user's question belongs to the anti-counterfeiting solution consultation category, the customer service large model assistant collects user information and determines whether it is complete based on the collected user information. If it is incomplete, the customer service large model assistant provides information guidance and guides the user to provide complete user information. If the user provides complete information, the case technical knowledge base is retrieved according to the user's current question and user information, and the product manager large model assistant generates a corresponding solution based on the retrieved content; when the user's question belongs to other questions, the customer service large model analyzes the problem and guides the user to ask questions related to anti-counterfeiting. Finally, the results of solving each type of problem are returned to the user.

[0106] In this application, the collaborative working mechanism of the multi-role large model assistant is utilized, and multiple specialized large model assistants (such as the technical customer service large model assistant and the product manager large model assistant) are used to handle different types of problems. These assistants can work together to provide targeted solutions. In this application, the intelligent customer service system can provide customers with efficient and accurate anti-counterfeiting technical consultation and anti-counterfeiting solution generation services. The system realizes the intelligent processing of anti-counterfeiting problems through the steps of automated context processing, customer information collection and guidance, knowledge base query and customized large model analysis, and significantly improves service efficiency and accuracy. By improving service efficiency and accuracy, this application can not only significantly reduce customer service costs, but also further enhance customer satisfaction and enhance corporate competitiveness.

[0107] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0108] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto. The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification. This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. An intelligent question-answering system based on a large language model, characterized in that: The intelligent question-answering system based on the large language model comprises: User question acquisition module, used to obtain the user's current question; A context processing module, connected to the user question acquisition module, for using a large language model to generate a search statement based on the user's historical question information and the user's current question; A question classification module, connected to the context processing module and the user question acquisition module, is used to classify the user's current question according to the search statement to obtain a variety of question classification results; the question classification results include: technical consulting questions, solution consulting questions, and questions other than technical consulting questions and solution consulting questions; A knowledge base management module is used to store technical solutions; the technical solutions include: technical knowledge content and application cases of the technical knowledge content; A question-answer processing module, connected to the user question acquisition module, the question classification module and the knowledge base management module, for retrieving the technical solution stored in the knowledge base management module based on the question classification result; The reply module is connected to the question-answer processing module and is used to adopt the large language model to generate reply content based on the retrieved technical solution.

2. The intelligent question-answering system based on a large language model according to claim 1, characterized in that: Also includes: The visualization module is connected to the reply module and is used to visualize and display the reply content.

3. The intelligent question-answering system based on a large language model according to claim 1, characterized in that: The context processing module comprises: User history question query unit, used to query user history question information; The search statement generation unit is connected to the user question acquisition module and the user history question query unit, and is used to use the large language model to generate a search statement carrying context information based on the user history question information and the user's current question.

4. The intelligent question-answering system based on a large language model according to claim 1, characterized in that: The knowledge base management module includes: A data storage unit, used to store the technical solution; An updating and maintaining unit is connected to the data storage unit and is used to update and maintain the technical solution stored in the data storage unit.

5. The intelligent question-answering system based on a large language model according to claim 1, characterized in that: The question-answer processing module comprises: A technical consultation question processing unit, connected to the user question acquisition module, the question classification module and the knowledge base management module, for retrieving the technical knowledge content stored in the knowledge base management module according to the technical consultation question; A solution consultation question processing unit, connected to the user question acquisition module, the question classification module and the knowledge base management module, for retrieving technical knowledge content and application cases of technical knowledge content stored in the knowledge base management module according to the solution consultation question; The question processing unit is connected with the user question acquisition module, the question classification module and the knowledge base management module, and is used to process questions other than technical consultation questions and solution consultation questions.

6. The intelligent question-answering system based on a large language model according to claim 5, characterized in that: The technical consultation problem processing unit includes: The keyword generation subunit is connected to the user question acquisition module and is used to generate search keywords based on the user's historical question information and the user's current question using a large language model; The technical search result generating subunit is connected with the question classification module, the keyword generating subunit and the knowledge base management module, and is used to retrieve the technical knowledge content stored in the knowledge base management module based on the search keyword.

7. The intelligent question-answering system based on a large language model according to claim 5, characterized in that: The solution consultation problem processing unit includes: An information collection subunit, used to collect user information; A judgment subunit, connected to the information collection subunit, for judging the integrity of the user information; The solution retrieval result generation subunit is connected to the user question acquisition module, the question classification module, the information collection subunit, the knowledge base management module and the judgment subunit, and is used to use a large language model to retrieve the technical knowledge content and technical knowledge content application cases stored in the knowledge base management module based on the user information and the solution consultation question.

8. The intelligent question-answering system based on a large language model according to claim 5, characterized in that: The problem processing unit comprises: The language guidance subunit is connected to the user question acquisition module and the question classification module, and is used to use a large language model to generate guidance information based on questions other than technical consultation questions and solution consultation questions.

9. The intelligent question answering system based on a large language model according to claim 1, characterized in that: Also includes: The feedback module is connected to the reply module and is used to obtain user usage information and provide feedback on the reply content based on the user usage information.

10. An intelligent question answering method based on a large language model, characterized in that: The intelligent question answering method based on the large language model includes: Get the user's historical question information and the user's current question; Using a large language model, a search statement is generated according to the user's historical question information and the user's current question; Classify the user's current question according to the search statement to obtain multiple question classification results; the question classification results include: technical consulting questions, solution consulting questions, and questions other than technical consulting questions and solution consulting questions; Retrieving the technical solutions stored in the knowledge base based on the problem classification result; The large language model is used to generate reply content based on the retrieved technical solution.

Images