An intelligent question and answer method and device, an electronic device, and a storage medium

Abstract

The application discloses an intelligent question and answer method and device, electronic equipment and storage medium, and relates to the technical field of artificial intelligence. The method comprises the following steps: obtaining an original input question of a user, and performing matching search on the original input question in a pre-constructed professional term explanation library to obtain a target professional term matched with the original input question and an explanation of the target professional term; performing rewriting processing on the original input question according to the target professional term and the explanation of the target professional term to obtain a target input question; performing knowledge search in a pre-constructed knowledge base according to the target input question, and generating reply content for the user according to a search result. The original input question of the user is rewritten based on the professional term and the explanation, so that the rewritten input question and the text in the knowledge base are more likely to have co-occurrence words, so that the knowledge search effect can be improved, and the accuracy of the reply content generated by the intelligent question and answer can be ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, and in particular to an intelligent question answering method and device, electronic equipment and a storage medium. BACKGROUND

[0002] A large language model refers to a natural language processing model with a large scale and a large number of parameters trained using deep learning technology. Such a model is usually based on a neural network architecture, which includes tens of billions or even hundreds of billions of parameters, enabling it to learn and understand complex structures and semantic relationships in natural language. The training of a large language model usually relies on a large-scale text corpus to improve the model's understanding and generation capabilities of language. In recent years, large language models have been widely used in question answering systems, in which relevant knowledge is recalled based on keywords included in the user's original input question, and then the large language model generates a reply content based on the recalled relevant knowledge. However, this type of question answering system has certain deficiencies: in a vertical field, the user's input question is relatively casual, and the keywords included therein cannot accurately recall the key knowledge, resulting in inaccurate final reply content. SUMMARY

[0003] The present application provides an intelligent question answering method, device, electronic equipment and storage medium.

[0004] According to another aspect of the present application, an intelligent question answering method is provided, comprising:

[0005] Obtaining an original input question of a user, and performing matching retrieval on the original input question in a pre-constructed professional term explanation library to obtain a target professional term matched with the original input question and an explanation of the target professional term;

[0006] Rewriting the original input question according to the target professional term and the explanation of the target professional term to obtain a target input question;

[0007] Performing knowledge retrieval on the target input question in a pre-constructed knowledge base, and generating a reply content for the user according to the retrieval result; wherein the professional term explanation library and the knowledge base are both constructed based on professional documents of a vertical field.

[0008] Optionally, the process of constructing the professional term explanation library comprises:

[0009] Performing document parsing processing and document segmentation processing on the professional documents of the vertical field to obtain a plurality of knowledge texts;

[0010] extracting professional terms from each of the knowledge texts, and interpreting the professional terms according to the context associated with the professional terms, to obtain a plurality of the professional terms and the interpretation of each of the professional terms;

[0011] constructing the professional term interpretation library according to the extracted professional terms and the interpretation of the professional terms.

[0012] Optionally, the extracting professional terms from each of the knowledge texts, and interpreting the professional terms according to the context associated with the professional terms, to obtain a plurality of the professional terms and the interpretation of each of the professional terms, comprises:

[0013] obtaining a pre-constructed language prompt; wherein the language prompt is used to guide the large language model to understand and process the knowledge texts;

[0014] inputting the knowledge texts and the language prompt into the large language model, so that the large language model extracts professional terms from the knowledge texts according to the language prompt, and interprets the professional terms according to the context of the professional terms, and outputs a plurality of professional terms and the interpretation of each of the professional terms.

[0015] Optionally, the constructing the professional term interpretation library according to the extracted professional terms and the interpretation of the professional terms comprises:

[0016] constructing an inverted index according to the professional terms and the interpretation of the professional terms; or, vectorizing the interpretation of the professional terms, and constructing a vector index between the professional terms and the interpretation of the professional terms;

[0017] performing database processing on the professional terms and the interpretation of the professional terms after the index is constructed, to obtain a professional term interpretation library.

[0018] Optionally, the rewriting the original input question according to the target professional term and the interpretation of the target professional term to obtain a target input question comprises:

[0019] inputting the target professional term, the interpretation of the target professional term, and the original input question into a pre-trained question rewriting model, so that the question rewriting model rewrites the original input question according to the target professional term and the interpretation of the target professional term, to obtain a rewritten target input question.

[0020] Optionally, the process of constructing the knowledge base comprises:

[0021] performing document parsing processing and document segmentation processing on the professional documents of the vertical field, to obtain a plurality of knowledge texts;

[0022] The plurality of pieces of knowledge text are converted into text vectors by embedding a model, and each piece of knowledge text and a text vector corresponding to the knowledge text are stored in a database, to obtain the knowledge base.

[0023] According to another aspect of the present application, an intelligent question answering device is provided, comprising:

[0024] A professional term retrieval module is configured to acquire an original input question of a user, and perform matching retrieval on the original input question in a pre-constructed professional term explanation library, to obtain a target professional term matched with the original input question and an explanation of the target professional term.

[0025] A question rewriting module is configured to perform rewriting processing on the original input question according to the target professional term and the explanation of the target professional term, to obtain a target input question.

[0026] A knowledge retrieval and reply module is configured to perform knowledge retrieval in a pre-constructed knowledge base according to the target input question, and generate reply content for the user according to a retrieval result. The professional term explanation library and the knowledge base are both constructed based on professional documents of a vertical field.

[0027] Shouldn't there be a professional vocabulary construction module here? This module and the professional vocabulary retrieval module together form part of the overall scheme.

[0028] Optionally, the intelligent question answering device further comprises:

[0029] A text processing module is configured to perform document parsing processing and document segmentation processing on the professional documents of the vertical field, to obtain a plurality of pieces of knowledge text.

[0030] A professional explanation extraction module is configured to extract professional terms from each piece of knowledge text, and explain the professional terms according to the context associated with the professional terms, to obtain a plurality of professional terms and an explanation of each professional term.

[0031] A library construction module is configured to construct the professional term explanation library according to the extracted professional terms and the explanations of the professional terms.

[0032] Optionally, the professional explanation extraction module is specifically configured to:

[0033] acquire a pre-constructed language prompt. The language prompt is used to guide a large language model to understand and process the knowledge text.

[0034] The knowledge text and the language prompts are input into the large language model, which extracts professional terms from the knowledge text based on the language prompts, and interprets the professional terms based on their context, outputting multiple professional terms and their explanations.

[0035] Optionally, the database creation module is specifically used for:

[0036] Based on the technical terms and their explanations, construct an inverted index; or, vectorize the explanations of the technical terms and construct a vector index between the technical terms and their explanations.

[0037] The indexed technical terms and their explanations are then stored in a database to obtain a technical term explanation database.

[0038] Optionally, the problem rewriting module 402 is specifically used for:

[0039] The target technical terms, their explanations, and the original input question are input into a pre-trained question rewriting model. The model then rewrites the original input question based on the target technical terms and their explanations, resulting in the rewritten target input question.

[0040] Optionally, a knowledge base building module is also included, used for:

[0041] The professional documents in the vertical field are parsed and segmented to obtain multiple knowledge texts.

[0042] The multiple knowledge texts are converted into text vectors using an embedding model, and each knowledge text and its corresponding text vector are stored in a database to obtain the knowledge base.

[0043] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0044] At least one processor; and

[0045] A memory that is communicatively connected to at least one processor; wherein,

[0046] The memory stores a computer program that can be executed by at least one processor, such that the at least one processor is able to execute the intelligent question-answering method of the present invention.

[0047] According to another aspect of the present invention, a computer-readable storage medium is provided, which stores computer instructions for causing a processor to execute and implement the intelligent question-answering method of the embodiments of the present invention.

[0048] The technical solution of this invention pre-constructs a professional terminology explanation library. Then, during the intelligent question-answering process, it first determines the professional terms and their explanations that match the user's original input question. Then, it rewrites the user's original input question based on the determined professional terms and their explanations. In this way, the rewritten user input question is closer to the document description in the knowledge base, and there are co-occurring words between the rewritten user input question and the knowledge document, making it easier to retrieve relevant knowledge results, thereby improving the retrieval effect and enabling the large model to generate more accurate response content.

[0049] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This is a flowchart illustrating an intelligent question-answering method according to an embodiment of the present invention;

[0052] Figure 2 This is a flowchart illustrating another intelligent question-answering method provided according to an embodiment of the present invention;

[0053] Figure 3 This is a logical schematic diagram of an intelligent question-answering method provided according to an embodiment of the present invention;

[0054] Figure 4 This is a schematic diagram of the structure of an intelligent question-answering device according to an embodiment of the present invention;

[0055] Figure 5 This is a schematic diagram of the structure of an electronic device that implements the intelligent question-answering method of this invention. Detailed Implementation

[0056] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0057] Example 1

[0058] Figure 1 The present invention provides a flowchart of an intelligent question-answering method. This embodiment is applicable to scenarios of intelligent question answering based on a large language model, typically applicable to intelligent question answering scenarios in vertical fields. The method can be executed by an intelligent question-answering device, which can be implemented in hardware and / or software and can be configured in an electronic device.

[0059] like Figure 1 As shown, this intelligent question-answering method includes:

[0060] S101. Obtain the user's original input question, and perform a matching search on the original input question in a pre-built professional term explanation library to obtain the target professional term that matches the original input question and the explanation of the target professional term.

[0061] In this embodiment of the invention, the professional terminology explanation library is pre-built based on professional documents in a vertical field. The library includes professional terms extracted from these documents and their explanations determined according to their context. A vertical field, in the internet industry, refers to a niche market within a specific industry, market, or demand area that focuses on providing in-depth information and professional services. This niche contrasts with generalized, comprehensive services or products, emphasizing professionalism, depth, and unique value within that field. For example, a vertical field could be the medical field, the financial field, or the insurance field. Professional documents in a vertical field refer to detailed, in-depth, and professional documents or materials written specifically for a particular industry, field, or professional need. These documents typically contain professional knowledge, technical details, operating guidelines, standards and specifications, case studies, etc., aiming to provide users with comprehensive, accurate, and practical information support.

[0062] Based on the completed professional terminology explanation database, when performing intelligent question answering, the intelligent question answering system based on a large language model first obtains the user's original input question. The original input question can be in text or speech format, without specific limitations. For the user's original input question, the similarity between the original input question and the explanations of each professional term in the professional terminology explanation database is calculated to determine the target professional term and its explanation. It is understood that the explanations of professional terms in the professional terminology explanation database can be stored as text vectors. Therefore, to calculate the similarity between the original input question and the explanations of each professional term in the database, the original input question needs to be converted into a text vector, and then the similarity is determined by calculating the distance between the text vectors. Converting the original input question into a corresponding text vector includes: if the original input question is in text format, it can be converted into a text vector using an embedding model; if the original input question is in speech format, it can be converted into text using speech conversion technology, and then converted into a text vector using an embedding model. In this embodiment, a similarity threshold can be preset, and professional terms with a calculated similarity exceeding the threshold are considered as the target professional terms. It should be noted that if there are multiple target professional terms, you can directly use all of them together to rewrite the original input question, or you can sort the target professional terms according to their similarity and select the target professional term with the highest similarity to use for subsequent rewriting of the original input question.

[0063] S102. Based on the target technical terms and their explanations, the original input problem is rewritten to obtain the target input problem.

[0064] In this embodiment of the invention, when rewriting the original input question, it can be first determined whether the explanation of the target technical term is related to the original input question, for example, by semantic analysis; if it is related, the original input question is rewritten based on the target technical term, for example, by replacing the keywords in the original input question with the target technical term, or by supplementing the original input question with the target technical term, so that the rewritten target input question can be a question that includes the target technical term.

[0065] In one optional implementation, a question rewriting model can be pre-trained. This pre-trained model can be a large language model or a separately trained general rewriting model; no specific limitation is made here. Based on this, the target technical term, its explanation, and the original input question are input into the pre-trained question rewriting model. The model then rewrites the original input question based on the target technical term and its explanation, resulting in the rewritten target input question.

[0066] Understandably, the rewritten target input question includes specialized terms from the vertical field. During subsequent knowledge retrieval, these specialized terms can be used to accurately retrieve key knowledge from the vertical field, thereby improving the effectiveness of knowledge retrieval.

[0067] For example, the user's original input question is: "How much do I need to spend before I get reimbursed?" The target technical term retrieved from the technical terminology database based on this original input question is "deductible." The definition of the target technical term is: "Deductible: In insurance, the amount that the insurance company will not pay. In the event of an insured event, if the loss is less than or equal to the deductible, the insurance company will not pay; if the loss exceeds the deductible, the insurance company will only pay the amount exceeding the deductible." The rewritten target input user question based on the target technical term and its definition could be: "What is my deductible?"

[0068] S103. Based on the target input question, perform knowledge retrieval in a pre-built knowledge base, and generate response content for the user based on the retrieval results.

[0069] In this embodiment of the invention, the knowledge base is constructed based on professional documents in a vertical field. The construction process may include: performing document parsing processing (e.g., layout analysis and information extraction processing) and document segmentation processing on the professional documents in the vertical field to obtain multiple knowledge texts. It should be noted that the reason for performing document segmentation processing is to ensure that the number of words in each knowledge text is within a preset limit. Furthermore, the multiple knowledge texts are converted into text vectors through an embedding model, and each knowledge text and its corresponding text vector are stored in a database to obtain the knowledge base.

[0070] After obtaining the target input question through step S102, a search can be performed directly in the knowledge base based on the target input question. For example, keywords can be extracted from the target input question to obtain multiple keywords, including target professional terms. Then, a search can be performed in the knowledge base based on these multiple keywords to obtain at least the knowledge related to the target professional terms included in the target input question (i.e., search results). The large language model used for intelligent question answering can then generate response content based on the obtained knowledge related to the target professional terms and provide it to the user, thus completing a question-and-answer session.

[0071] In this embodiment of the invention, a professional terminology explanation library is pre-built. Then, during the intelligent question-answering process, the target professional terms and their explanations corresponding to the user's original input question are first determined. The user's original input question is then rewritten based on the determined target professional terms and their explanations. In this way, the rewritten user input question is closer to the document description in the knowledge base. That is, there are more likely to be co-occurring words between the rewritten user input question and the knowledge documents in the knowledge base, making it easier to retrieve relevant knowledge results and thus improving the accuracy of retrieval, so that the large model can generate more accurate response content.

[0072] Example 2

[0073] Figure 2 A flowchart of an intelligent question-answering method is provided for this embodiment of the invention. For the specific process of constructing a specialized terminology explanation database added in this embodiment, please refer to [link to documentation]. Figure 2 The method includes the following steps:

[0074] S201. Perform document parsing and document segmentation on the professional documents in the vertical field to obtain multiple knowledge texts.

[0075] In this embodiment of the invention, document parsing processing may include layout analysis processing and information extraction processing. Layout analysis processing may include: first, format conversion, primarily converting the original format of the professional document (e.g., PDF or Word) into a format easier for computers to process, such as an image format; then, layout structure recognition, using deep learning models such as convolutional neural networks (CNNs) to analyze the document's layout structure. These models can identify different areas in the document, such as titles, body text, images, tables, etc.; finally, layout feature extraction, such as extracting the spatial relationships between different blocks in the document, such as position, size, and arrangement order. Information extraction processing may include: extracting the document's title from the segmented title area; for paragraph areas, reading and organizing line by line to identify key content within paragraphs; for table areas, identifying table borders and cell content, and converting the table content into structured data. It is understood that the above processing is merely illustrative and may include other processing methods, which are not listed here. Document parsing of professional documents involves integrating extracted information to obtain structured text content. This text content can then be segmented, for example, by using a fixed text length as a segmentation criterion. Segmentation occurs when the accumulated text length reaches a set value. Alternatively, semantic analysis can be performed on the document content, leading to segmentation based on semantic analysis. Ultimately, multiple pieces of knowledge text are obtained.

[0076] S202. Extract professional terms from each piece of knowledge text, and interpret the professional terms according to the context associated with them, to obtain multiple professional terms and the interpretation of each professional term.

[0077] In one optional implementation, for a knowledge text, the words it contains are converted into word vectors using word embedding. Then, a clustering algorithm groups similar word vectors, with each group potentially representing a technical term or concept. Alternatively, word frequency statistics can be used to determine the technical terms included in each knowledge text. For each extracted technical term, the specific location of the term in the knowledge text is found, identifying the sentence or paragraph in which it appears. This is the context associated with the technical term, i.e., its linguistic context. The meaning and function of the technical term in the context are analyzed. Considering both the literal meaning of the technical term and its specific meaning in a particular vertical field, the following methods can be used to explain the term: Definition method: Provide a professional definition of the term, explaining its precise meaning in that field. Example method: Use specific examples to illustrate the meaning of the term, making the explanation more intuitive. Comparison method: Compare the term with similar terms to highlight its characteristics and differences.

[0078] In another alternative implementation, a pre-built language prompt can be obtained. This prompt guides the large language model in understanding and processing the knowledge text. The knowledge text and the prompt are input into the large language model (a pre-trained model for extracting specialized terms and their explanations). The large language model then extracts specialized terms from the knowledge text based on the prompt and explains these terms according to their context, outputting multiple specialized terms and their explanations. It is understood that using a large model to extract specialized terms and their explanations can improve the efficiency and accuracy of this process.

[0079] S203. Based on the extracted professional terms and their explanations, construct the professional term explanation database.

[0080] In one optional implementation, an inverted index is constructed based on the technical terms and their explanations; or, the explanations of the technical terms are vectorized, and a vector index between the technical terms and their explanations is constructed; the indexed technical terms and their explanations are then stored in a database to obtain a technical term explanation database.

[0081] S204. Obtain the user's original input question, and perform a matching search on the original input question in a pre-built professional term explanation library to obtain the target professional term that matches the original input question and the explanation of the target professional term.

[0082] S205. Based on the target technical terms and their explanations, the original input problem is rewritten to obtain the target input problem.

[0083] S206. Based on the target input question, perform knowledge retrieval in a pre-built knowledge base, and generate response content for the user based on the retrieval results.

[0084] In this embodiment, professional terms and their explanations can be extracted from professional documents in a vertical field to build a professional term explanation library. This is a prerequisite for rewriting the user's original input question, that is, building a professional term explanation library provides a foundation for accurately rewriting the user's original input question.

[0085] Example 3

[0086] Figure 3 A logic diagram of an intelligent question-answering method is provided for an embodiment of the present invention. See also... Figure 3 The method includes the following steps:

[0087] First, professional documents in the vertical field are acquired. These documents are then parsed and segmented. Based on the segmentation results, two paths are taken: one path constructs knowledge based on the segmentation results (e.g., embedding the segmented documents into a database) to obtain a knowledge base; the other path extracts professional terms and their explanations based on the segmentation results. The specific extraction process is described in the above embodiment and will not be repeated here. Finally, a professional term explanation database is constructed based on the extracted professional terms and their explanations. This is the preprocessing step.

[0088] The intelligent question-answering process includes: obtaining the user's original input question; retrieving professional terms and their explanations that match the original input question based on a professional terminology explanation database (i.e., professional terminology retrieval in the diagram); finally, through a rewriting module (which can be a large language model specifically for question rewriting or a general model), rewriting the question based on the original input question, the retrieved target professional terms and their explanations, to obtain a target input question that includes professional terms; based on the target input question, performing knowledge retrieval from the knowledge base; and finally generating a response based on the retrieval results and providing it back to the user, thus concluding one question-answering session.

[0089] In this embodiment of the invention, a professional terminology explanation library is pre-built. Then, during the intelligent question-answering process, the target professional terms and their explanations corresponding to the user's original input question are first determined. The user's original input question is then rewritten based on the determined target professional terms and their explanations. In this way, the rewritten user input question is closer to the document description in the knowledge base. That is, there are more likely to be co-occurring words between the rewritten user input question and the knowledge documents in the knowledge base, making it easier to retrieve relevant knowledge results and thus improving the accuracy of retrieval, so that the large model can generate more accurate response content.

[0090] Example 4

[0091] Figure 4 This is a schematic diagram of the structure of an intelligent question-answering device provided in an embodiment of the present invention. This embodiment is applicable to scenarios of intelligent question answering based on large language models, typically suitable for intelligent question answering scenarios in vertical fields. Figure 4 As shown, the device includes:

[0092] The professional term retrieval module 401 is used to obtain the user's original input question and perform matching retrieval on the original input question in a pre-built professional term explanation library to obtain the target professional term that matches the original input question and the explanation of the target professional term;

[0093] The question rewriting module 402 is used to rewrite the original input question based on the target technical terms and their explanations to obtain the target input question.

[0094] The knowledge retrieval and response module 403 is used to perform knowledge retrieval in a pre-built knowledge base based on the target input question, and generate response content for the user based on the retrieval results; wherein, the professional terminology explanation library and the knowledge base are both built based on professional documents in the vertical field.

[0095] In one alternative implementation, it also includes:

[0096] The text processing module is used to perform document parsing and document segmentation on professional documents in the vertical field to obtain multiple knowledge texts;

[0097] The professional explanation extraction module is used to extract professional terms from each piece of knowledge text and interpret the professional terms according to the context associated with them, thereby obtaining multiple professional terms and the interpretation of each professional term.

[0098] The database construction module is used to construct the professional term explanation database based on the extracted professional terms and their explanations.

[0099] In one alternative implementation, the professional interpretation extraction module is specifically used for:

[0100] Obtain pre-constructed language prompts; wherein the language prompts are used to guide the large language model to understand and process the knowledge text;

[0101] The knowledge text and the language prompts are input into the large language model, which extracts professional terms from the knowledge text based on the language prompts, and interprets the professional terms based on their context, outputting multiple professional terms and their explanations.

[0102] In one optional implementation, the library creation module is specifically used for:

[0103] Based on the technical terms and their explanations, construct an inverted index; or, vectorize the explanations of the technical terms and construct a vector index between the technical terms and their explanations.

[0104] The indexed technical terms and their explanations are then stored in a database to obtain a technical term explanation database.

[0105] In one alternative implementation, the problem rewriting module 402 is specifically used for:

[0106] The target technical terms, their explanations, and the original input question are input into a pre-trained question rewriting model. The model then rewrites the original input question based on the target technical terms and their explanations, resulting in the rewritten target input question.

[0107] In one alternative implementation, a knowledge base building module is also included, used for:

[0108] The professional documents in the vertical field are parsed and segmented to obtain multiple knowledge texts.

[0109] The multiple knowledge texts are converted into text vectors using an embedding model, and each knowledge text and its corresponding text vector are stored in a database to obtain the knowledge base.

[0110] The intelligent question-answering device provided in the embodiments of the present invention can execute the intelligent question-answering method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the method execution.

[0111] Example 5

[0112] Figure 5 A schematic diagram of an electronic device 10 that can be used to implement embodiments of the present invention is shown. The electronic device is intended optionally as a controller device in a new energy vehicle. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0113] like Figure 5 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0114] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disks, optical disks, etc.; and communication unit 19, such as network cards, modems, wireless transceivers, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0115] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as performing intelligent question answering methods.

[0116] In some embodiments, the intelligent question-answering method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the intelligent question-answering method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to execute the intelligent question-answering method by any other suitable means (e.g., by means of firmware).

[0117] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0118] Computer programs used to implement the methods of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to the processor of a general-purpose computer, a special-purpose computer, or other programmable intelligent question-answering device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The computer programs can be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0119] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0120] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0121] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0122] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0123] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0124] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. An intelligent question-answering method, characterized in that, include: Perform document parsing and segmentation on professional documents in a vertical field to obtain multiple knowledge texts; Obtain pre-constructed language prompts; wherein the language prompts are used to guide the large language model to understand and process the knowledge text; The knowledge text and the language prompts are input into the large language model, which extracts professional terms from the knowledge text based on the language prompts, and interprets the professional terms based on their context, outputting multiple professional terms and their explanations. Based on the extracted professional terms and their explanations, construct the professional term explanation database; The system obtains the user's original input question and performs a matching search on the original input question in a pre-built professional term explanation database to obtain the target professional term that matches the original input question and the explanation of the target professional term; wherein, the original input question refers to the question text or question voice input by the user based on the big language intelligent question answering system; The target technical terms, their explanations, and the original input question are input into a pre-trained question rewriting model. The model then rewrites the original input question based on the target technical terms and their explanations, resulting in a rewritten target input question. The question rewriting model is a large language model. The system retrieves knowledge from a pre-built knowledge base based on the target input question and generates a response for the user based on the retrieval results; wherein, both the professional terminology explanation database and the knowledge base are built based on professional documents in a vertical field.

2. The method according to claim 1, characterized in that, The step of constructing the professional term explanation database based on the extracted professional terms and their explanations includes: Based on the technical terms and their explanations, construct an inverted index; or, vectorize the explanations of the technical terms and construct a vector index between the technical terms and their explanations. The indexed technical terms and their explanations are then stored in a database to obtain a technical term explanation database.

3. The method according to claim 1, characterized in that, The process of constructing the knowledge base includes: The professional documents in the vertical field are parsed and segmented to obtain multiple knowledge texts. The multiple knowledge texts are converted into text vectors using an embedding model, and each knowledge text and its corresponding text vector are stored in a database to obtain the knowledge base.

4. An intelligent question-and-answer device, characterized in that, include: The text processing module is used to parse and segment professional documents in a vertical field to obtain multiple knowledge texts. A professional explanation extraction module is used to obtain pre-constructed language prompts; wherein, the language prompts are used to guide the large language model to understand and process the knowledge text; the knowledge text and the language prompts are input into the large language model, so that the large language model extracts professional terms from the knowledge text according to the language prompts, and explains the professional terms according to the context of the professional terms, and outputs multiple professional terms and the explanation of each professional term; The database construction module is used to construct the professional term explanation database based on the extracted professional terms and their explanations. The professional term retrieval module is used to obtain the user's original input question and match and retrieve the original input question in a pre-built professional term explanation database to obtain the target professional term that matches the original input question and the explanation of the target professional term; wherein, the original input question refers to the question text or question voice input by the user based on the big language intelligent question answering system; The question rewriting module is used to input the target technical term, the explanation of the target technical term, and the original input question into a pre-trained question rewriting model, so that the question rewriting model rewrites the original input question according to the target technical term and the explanation of the target technical term to obtain the rewritten target input question; the question rewriting model is a large language model; The knowledge retrieval and response module is used to perform knowledge retrieval in a pre-built knowledge base based on the target input question, and generate response content for the user based on the retrieval results; wherein, the professional terminology explanation library and the knowledge base are both built based on professional documents in the vertical field.

5. An electronic device, characterized in that, include: At least one processor; as well as 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 to enable the at least one processor to perform the method of any one of claims 1-3.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that are used to cause a processor to execute the method of any one of claims 1-3.

Images