Intelligent question and answer method, device and equipment in electronic cigarette industry and storage medium

Abstract

The invention discloses an electronic cigarette industry intelligent question and answer method, device and equipment and a storage medium, and the electronic cigarette industry intelligent question and answer method comprises the steps: collecting and integrating electronic cigarette industry knowledge, and constructing a knowledge graph; training a large language model by using the knowledge graph to obtain an electronic cigarette industry large model; enriching and updating the knowledge graph by using the electronic cigarette industry large model; obtaining a first answer corresponding to a user question according to the electronic cigarette industry large model; comparing and correcting the first answer according to the knowledge graph to obtain a user answer; and optimizing the electronic cigarette industry large model according to the enriched and updated knowledge graph. The intelligent question-answering method in the electronic cigarette industry solves the technical problem that in the prior art, a user is lack of deep understanding of electronic cigarette products.

Description

Technical Field

[0001] The present application relates to the field of intelligent question answering, and in particular to an intelligent question answering method, device, equipment and storage medium in the electronic cigarette industry. Background Art

[0002] The speed of electronic cigarette product replacement is very fast, and new products with various new models and functions continue to emerge on the market, making it difficult for users to keep up with the pace of change. Due to the lack of in-depth understanding of these new products, many users are confused when choosing the right product for them and do not know how to make wise decisions. Summary of the invention

[0003] The main purpose of this application is to propose an intelligent question-answering method, device, equipment and storage medium for the electronic cigarette industry, which can solve the technical problem in the prior art that users lack in-depth understanding of electronic cigarette products.

[0004] To achieve the above objectives, the first aspect of the present application provides an intelligent question-answering method for the electronic cigarette industry, comprising:

[0005] Collect and integrate e-cigarette industry knowledge and build a knowledge graph;

[0006] Using the knowledge graph to train a large language model to obtain a large model of the e-cigarette industry;

[0007] Enriching and updating the knowledge graph using the electronic cigarette industry big model;

[0008] Obtaining a first answer corresponding to the user's question according to the electronic cigarette industry macro model;

[0009] Compare and correct the first answer according to the knowledge graph to obtain a user answer;

[0010] The electronic cigarette industry big model is optimized based on the enriched and updated knowledge graph.

[0011] Furthermore, the method of using the knowledge graph to train the large language model to obtain the large model of the electronic cigarette industry includes:

[0012] Extracting triple data from the knowledge graph;

[0013] Design a first question-answer pair based on the triple data in the knowledge graph;

[0014] The large language model is trained using the first question-answer pair as training data to obtain a large model of the electronic cigarette industry.

[0015] Furthermore, the step of using the first question-answer pair as training data to train the large language model to obtain a large model for the electronic cigarette industry includes: using the first question-answer pair to perform instruction fine-tuning on the large language model.

[0016] Further, the information generated by the electronic cigarette industry big model includes: structured data generated by using the electronic cigarette industry big model, and / or, natural language feedback generated by using the electronic cigarette industry big model;

[0017] The method of enriching and updating the knowledge graph by using the electronic cigarette industry big model includes:

[0018] Participate in the process of reconstructing the knowledge graph using the information generated by the electronic cigarette industry big model, and / or

[0019] The information generated by the electronic cigarette industry big model is used to participate in the process of generating the knowledge graph data into natural language.

[0020] Furthermore, the enriching and updating of the knowledge graph by using the electronic cigarette industry big model includes: using the embedding vectors of the electronic cigarette industry big model to supplement the representation of entities and relationships in the knowledge graph.

[0021] Furthermore, comparing and correcting the first answer according to the knowledge graph to obtain the user answer includes:

[0022] Generate a knowledge graph answer for the user question based on the knowledge graph;

[0023] Calculate the similarity between the first answer and the knowledge graph answer;

[0024] In response to the similarity being greater than or equal to a preset similarity, taking the first answer as the user answer; or

[0025] In response to the similarity being less than a preset similarity, the first answer is corrected according to the knowledge graph answer to obtain a second answer, and the second answer is used as the user answer.

[0026] Furthermore, before optimizing the electronic cigarette industry big model according to the enriched and updated knowledge graph, the electronic cigarette industry intelligent question-answering method further includes:

[0027] A second question-answer pair is generated according to the user question and the user answer, and the knowledge graph is enriched and updated again using the second question-answer pair.

[0028] The second aspect of the present application provides an intelligent question-answering device for the electronic cigarette industry, comprising:

[0029] Knowledge graph construction module, used to collect and integrate e-cigarette industry knowledge and build a knowledge graph;

[0030] An industry information module, used to train a large language model using the knowledge graph to obtain a large model of the e-cigarette industry;

[0031] A knowledge graph enrichment and update module, used to enrich and update the knowledge graph using the electronic cigarette industry big model;

[0032] A first answer generation module, used to obtain a first answer corresponding to the user's question according to the electronic cigarette industry macro model;

[0033] A user answer generation module, used to compare and correct the first answer according to the knowledge graph to obtain a user answer;

[0034] The model optimization module is used to optimize the electronic cigarette industry big model according to the enriched and updated knowledge graph.

[0035] Furthermore, the device also includes:

[0036] The knowledge graph is enriched and updated again by a module, which is used to generate a second question-answer pair according to the user question and the user answer, and to enrich and update the knowledge graph again by using the second question-answer pair.

[0037] The third aspect of the present application provides an intelligent question-answering device for the electronic cigarette industry, the device comprising a processor and a memory:

[0038] The memory is used to store program code and transmit the program code to the processor;

[0039] The processor is used to execute the electronic cigarette industry intelligent question-answering method described in any one of the first aspects according to the instructions in the program code.

[0040] A fourth aspect of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the electronic cigarette industry intelligent question-and-answer method described in any one of the first aspects above is implemented.

[0041] This application provides an intelligent question-answering method, device, equipment and storage medium for the electronic cigarette industry, which can provide users with timely and detailed product information, helping them better understand the characteristics, advantages and usage methods of different electronic cigarette products. Through this intelligent assistance, users can not only find products that suit them more easily, but also get a better user experience, effectively reducing the troubles and inconveniences caused by not understanding the products. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 An intelligent question-answering method for the electronic cigarette industry provided in an embodiment of the present application;

[0043] Figure 2 An intelligent question-and-answer device for the electronic cigarette industry provided in an embodiment of the present application;

[0044] Figure 3 An intelligent question-and-answer device for the electronic cigarette industry provided in an embodiment of the present application. DETAILED DESCRIPTION

[0045] In order to make the above-mentioned purpose, features and advantages of the present application more obvious and easy to understand, the specific implementation mode of the present application is described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present application, so the present application is not limited by the specific embodiments disclosed below. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present application. The terms used herein in the specification of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0046] For easier understanding, see Figure 1 , the present application provides an intelligent question-answering method for the electronic cigarette industry, the method comprising:

[0047] S100, collect and integrate e-cigarette industry knowledge and build a knowledge graph.

[0048] First, collect relevant knowledge about the e-cigarette industry from a variety of data sources. These data sources include industry standard documents, product manuals, market reports, user reviews, and product reviews. Use natural language processing technology to preprocess these data and identify the entities and relationships therein. For example, extract entities such as "e-cigarette device", "nicotine concentration", and "e-liquid ingredients" from the data, and identify the relationships between them, such as "e-cigarette device-contains-nicotine" or "nicotine concentration-affects-user experience". Organize these entities and their relationships into triples, and build a knowledge graph for the e-cigarette industry based on this.

[0049] S200: Use the knowledge graph to train a large language model to obtain a large model of the electronic cigarette industry.

[0050] In some embodiments, step S200 uses the knowledge graph to train a large language model to obtain a large model of the electronic cigarette industry, including:

[0051] S210, extracting triple data from the knowledge graph.

[0052] In this step, the method of the present application will obtain triple data from the knowledge graph. These triples consist of an entity, a relationship, and another related entity, forming knowledge links between entities. For example, a triple may include the connection between "electronic cigarette brand A" (entity 1) and "production" (relationship) and "a manufacturer" (entity 2). In this way, the question-answering method of the present application can extract structured information that is crucial to understanding the electronic cigarette industry.

[0053] S220. Design a first question-answer pair based on the triple data in the knowledge graph.

[0054] After the triple data is extracted, the data is used to generate the first question-answer pair. The designed first question-answer pair is not just a question and answer about simple facts, but also involves deep knowledge in the e-cigarette industry. For example, the corresponding first question-answer pair can be designed based on the question "Which manufacturer produces e-cigarette brand A?", or based on a more complex question "In the e-cigarette industry, what are the differences between the manufacturers of brand A and brand B?", and then the corresponding first question-answer pair can be designed based on the triple data. These first question-answer pairs cover multiple knowledge dimensions in the e-cigarette industry, ensuring the diversity and comprehensiveness of the training data.

[0055] S230: Use the first question-answer pair as training data to train the large language model to obtain a large model of the electronic cigarette industry.

[0056] After completing the design of the first question-answer pairs, the question-answering method of the present application can input these first question-answer pairs into the large language model as part of the training data. Through repeated training and optimization, the large language model can gradually learn and master the proprietary knowledge of the e-cigarette industry, and eventually form a large model specifically for the e-cigarette industry, that is, the e-cigarette industry large model. This model can not only more accurately understand and answer questions related to the e-cigarette industry, but also help users make more informed decisions in a complex information environment.

[0057] Furthermore, the question-answer pair is used as training data to train the large language model to obtain the large model of the electronic cigarette industry, and the large language model can also be trained using the first question-answer pair. Specifically, after completing the preliminary training and generating the large model of the electronic cigarette industry, the instruction fine-tuning technology can be further used to make more detailed adjustments to the model. Instruction fine-tuning enables the model to not only answer questions correctly, but also answer according to predetermined logic or industry standards by designing specific instructions and tasks.

[0058] During this process, the first question-answer pair will be used as the core data set for fine-tuning model instructions. By fine-tuning the model, it can perform more in line with the professional requirements of the e-cigarette industry in terms of understanding the context of the question, choosing the answer method, and providing detailed information. For example, when the model receives an instruction like "explain the difference in performance between e-cigarette products A and B", it must not only give basic facts, but also provide in-depth analysis and suggestions from the perspective of industry experts. In this way, the model after instruction fine-tuning can better handle complex e-cigarette industry-specific issues, thereby providing users with more valuable answers and insights.

[0059] S300: Enrich and update the knowledge graph using the electronic cigarette industry big model.

[0060] Specifically, there are three ways to enrich and update the knowledge graph using the electronic cigarette industry big model. The question-answering method of the present application can be arbitrarily combined based on these three methods according to actual needs. Two of the methods require first using the electronic cigarette industry big model to generate information, and this information includes at least one of structured data and natural language feedback. Furthermore, the first method can reconstruct the knowledge graph and use the information generated by the electronic cigarette industry big model to participate in the process of reconstructing the knowledge graph; the second method can use the information generated by the electronic cigarette industry big model to participate in the process of generating the knowledge graph data into natural language. The third method can use the embedding vector of the electronic cigarette industry big model to supplement the representation of entities and relationships in the knowledge graph.

[0061] The electronic cigarette industry big model is used to enrich and update the knowledge graph, and the electronic cigarette industry big model can be used to generate natural language text to describe the content of the knowledge graph or supplement the information in the knowledge graph. The electronic cigarette industry big model is used to perform complex logical reasoning, infer the implicit relationships or attributes that may exist in the knowledge graph, and enhance the connectivity and depth of the knowledge graph.

[0062] S400: Obtain a first answer corresponding to the user's question according to the electronic cigarette industry macro model.

[0063] S500: Compare and correct the first answer according to the knowledge graph to obtain a user answer.

[0064] In some embodiments, S500 compares and corrects the first answer according to the knowledge graph to obtain the user answer, including:

[0065] S510. Generate a knowledge graph answer for the user question based on the knowledge graph.

[0066] In the process of generating answers based on knowledge graphs, information retrieval is first performed. Specifically, relevant information is retrieved from the knowledge graph by identifying keywords and entities in the user's question. Then, in the answer construction phase, the retrieved information can be analyzed to determine which information is necessary to answer the question, and this information can be organized to form a logically coherent answer framework. Finally, natural language processing technology is used to convert this structured information into fluent, easy-to-understand natural language text to generate the final knowledge graph answer.

[0067] S520. Calculate the similarity between the first answer and the knowledge graph answer.

[0068] Specifically, in order to ensure the consistency between the first answer obtained according to the electronic cigarette industry model and the authoritative answer (i.e., the knowledge graph answer) already in the knowledge graph, a variety of similarity calculation methods can be used to evaluate the similarity between the two. For example, the method of the present application can use cosine similarity calculation. Cosine similarity determines the degree of similarity between two vectors by measuring the angle between them.

[0069] Furthermore, after calculating the similarity, the method of the present application will make different responses according to the value of the similarity.

[0070] In response to the similarity being greater than or equal to the preset similarity, the first answer is used as the user answer. Specifically, if the similarity is greater than or equal to the preset similarity, it means that the first answer and the knowledge graph answer have sufficient similarity in content and meaning. At this time, the question-answering method of the present application will directly use the first answer as the final answer provided to the user (i.e., the user answer). This processing method ensures the accuracy and reliability of the answer.

[0071] In response to the similarity being less than a preset similarity, the first answer is corrected according to the knowledge graph answer to obtain a second answer, and the second answer is used as the user answer. Specifically, if the similarity is less than the preset similarity, this indicates that the first answer may deviate significantly from the knowledge graph answer. To this end, the question-and-answer method of the present application may make necessary corrections and adjustments to the first answer based on the answer in the knowledge graph to generate a new answer, namely the second answer. Through this correction process, the second answer will be more accurate, closer to the facts, and able to better meet the needs of users. Finally, the question-and-answer method of the present application returns the second answer to the user as the user answer, ensuring that the information obtained by the user is verified and of high quality.

[0072] S600. Generate a second question-answer pair based on the user question and the user answer, and use the second question-answer pair to enrich and update the knowledge graph again.

[0073] Specifically, the method of the present application generates a new question-answer pair, namely the second question-answer pair, based on the user question and the final user answer. The purpose of this step is to record the valuable question-answer information generated during each user interaction process to form a new knowledge entry.

[0074] The process of generating the second question-answer pair is not a simple record, but by associating the question raised by the user with the user answer provided by the question-answering method of this application, the method of this application can form new triple data, thereby enhancing the content of the knowledge graph. For example, if a user asks a specific question about whether the smoking experience of electronic cigarette product A can be adjusted, the question-answering method of this application, after providing the user answer, pairs the user question with the user answer to generate a second question-answer pair, which can reveal new knowledge associations, or even details that have not yet been covered in the knowledge graph.

[0075] Next, the knowledge graph is enriched and updated again using the second question-answer pair. This means that the question-answering method of the present application can integrate the generated second question-answer pair into the existing knowledge graph as a new knowledge node or relationship to supplement and expand it. In this way, the knowledge graph can not only maintain the latest and accuracy of its information, but also gradually cover more complex information and emerging trends related to the e-cigarette industry. This updating process enables the knowledge graph to dynamically adapt to industry changes and improve the accuracy of the question-answering method of the present application in future questions and answers and its ability to deal with complex problems.

[0076] S700. Optimize the electronic cigarette industry big model according to the enriched and updated knowledge graph.

[0077] After enriching and updating the knowledge graph, the method of the present application can further optimize the electronic cigarette industry model based on the enriched and updated knowledge graph. Specifically, as the content of the knowledge graph continues to expand and improve, the industry knowledge it contains also grows, which provides a new data foundation and training materials for the optimization of the electronic cigarette industry model.

[0078] The method of the present application can comprehensively analyze the enriched and updated knowledge graph to identify newly added knowledge items and changed knowledge structures. This new information may include new industry trends, technological developments, market dynamics, new product evaluations, and changes in regulations and policies, or new questions raised by users and their related answers. These contents are introduced into the training process of the large model of the e-cigarette industry, so that the model can learn the latest industry knowledge and understand the relationship between this knowledge and existing knowledge.

[0079] Next, the question-answering method of the present application can use this rich and new information to retrain and fine-tune the large model of the e-cigarette industry. The core of this step is to ensure that the model can not only understand and process the latest information in the e-cigarette industry, but also improve its performance in question-answering tasks. For example, the method of the present application can perform reinforcement learning through the second question-answer pair and knowledge nodes, so that the model can more accurately predict the intention of the user's question and provide more precise and professional answers. At the same time, the method of the present application will also optimize the reasoning ability of the model, so that it can comprehensively consider multiple sources of knowledge and give the best answer when dealing with complex industry problems.

[0080] Through this continuous optimization process, the e-cigarette industry big model will become smarter and more powerful. It will not only keep pace with the times and reflect the latest changes in the industry, but also continuously improve its answer quality and ability to deal with complex questions, and ultimately provide users with higher quality intelligent question-and-answer services.

[0081] The electronic cigarette industry intelligent question-and-answer method provided in this application can provide users with timely and detailed product information through the above steps, helping them better understand the characteristics, advantages and usage methods of different electronic cigarette products. Through this intelligent assistance, users can not only find products that suit them more easily, but also get a better user experience, effectively reducing the troubles and inconveniences caused by not understanding the products.

[0082] Figure 2 An intelligent question-answering device for the electronic cigarette industry provided in this application includes:

[0083] The knowledge graph construction module 11 is used to collect and integrate the knowledge of the electronic cigarette industry and construct a knowledge graph;

[0084] The industry information module 12 is used to train the large language model using the knowledge graph to obtain a large model of the electronic cigarette industry;

[0085] A knowledge graph enrichment and update module 13, used to enrich and update the knowledge graph using the electronic cigarette industry big model;

[0086] A first answer generation module 14, used to obtain a first answer corresponding to the user's question according to the electronic cigarette industry macro model;

[0087] The user answer generation module 15 is used to compare and correct the first answer according to the knowledge graph to obtain the user answer.

[0088] The model optimization module 17 is used to optimize the electronic cigarette industry big model according to the enriched and updated knowledge graph.

[0089] The electronic cigarette industry intelligent question-and-answer device provided in this embodiment is used to execute the question-and-answer method provided in any of the aforementioned method embodiments. The electronic cigarette industry intelligent question-and-answer device can provide users with timely and detailed product information to help them better understand the characteristics, advantages and usage methods of different electronic cigarette products. Through this intelligent assistance, users can not only find products that suit them more easily, but also get a better user experience, effectively reducing the troubles and inconveniences caused by not understanding the products.

[0090] Furthermore, the electronic cigarette industry intelligent question-answering device also includes:

[0091] The knowledge graph re-enrichment and update module 16 is used to generate a second question-answer pair based on the user question and the user answer, and use the second question-answer pair to enrich and update the knowledge graph again.

[0092] By using the knowledge graph to enrich the update module 16 again, the knowledge graph can not only keep its information up to date and accurate, but also gradually cover more complex information and emerging trends related to the e-cigarette industry. This update process enables the knowledge graph to dynamically adapt to industry changes and improve the accuracy of the question-answering method of this application in future question-answering and its ability to deal with complex problems.

[0093] Figure 3 This application provides an electronic cigarette industry intelligent question-answering device, which includes a processor 21 and a memory 22:

[0094] The memory 22 is used to store program codes and transmit the program codes to the processor;

[0095] The processor 21 is used to execute the intelligent question-answering method for the electronic cigarette industry described in any of the above embodiments according to the instructions in the program code.

[0096] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0097] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0098] In the several embodiments provided in this application, it should be understood that the disclosed methods, devices, equipment and storage media can be implemented in other ways. For example, the system embodiments described above are only schematic, for example, the division of modules is only a logical function division, and there may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another device, or some features can be ignored or not executed.

[0099] In addition, each functional module in each embodiment of the present application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The above integrated modules can be implemented in the form of hardware or software functional modules.

[0100] If the integrated module is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program code.

[0101] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. An intelligent question-answering method for the electronic cigarette industry, characterized in that: include: Collect and integrate e-cigarette industry knowledge and build a knowledge graph; Using the knowledge graph to train a large language model to obtain a large model of the e-cigarette industry; Enriching and updating the knowledge graph using the electronic cigarette industry big model; Obtaining a first answer corresponding to the user's question according to the electronic cigarette industry macro model; Compare and correct the first answer according to the knowledge graph to obtain a user answer; The electronic cigarette industry big model is optimized based on the enriched and updated knowledge graph.

2. The electronic cigarette industry intelligent question-answering method according to claim 1, characterized in that: The method of using the knowledge graph to train the large language model to obtain the large model of the electronic cigarette industry includes: Extracting triple data in the knowledge graph; Design a first question-answer pair based on the triple data in the knowledge graph; The large language model is trained using the first question-answer pair as training data to obtain a large model of the electronic cigarette industry.

3. The intelligent question-answering method for the electronic cigarette industry according to claim 2, characterized in that: The method of using the first question-answer pair as training data to train the large language model to obtain a large model for the electronic cigarette industry includes: using the first question-answer pair to perform instruction fine-tuning on the large language model.

4. The intelligent question-answering method for the electronic cigarette industry according to claim 1, characterized in that: The information generated by the electronic cigarette industry big model includes: structured data generated by the electronic cigarette industry big model, and / or natural language feedback generated by the electronic cigarette industry big model; The method of enriching and updating the knowledge graph by using the electronic cigarette industry big model includes: Participate in the process of reconstructing the knowledge graph using the information generated by the electronic cigarette industry big model, and / or The information generated by the electronic cigarette industry big model is used to participate in the process of generating the knowledge graph data into natural language.

5. The intelligent question-answering method for the electronic cigarette industry according to claim 1 or 4, characterized in that: The enriching and updating of the knowledge graph by using the electronic cigarette industry big model includes: using the embedding vectors of the electronic cigarette industry big model to supplement the representation of entities and relationships in the knowledge graph.

6. The intelligent question-answering method for the electronic cigarette industry according to claim 1, characterized in that: The comparing and correcting the first answer according to the knowledge graph to obtain the user answer includes: Generate a knowledge graph answer for the user question based on the knowledge graph; Calculate the similarity between the first answer and the knowledge graph answer; In response to the similarity being greater than or equal to a preset similarity, taking the first answer as the user answer; or In response to the similarity being less than a preset similarity, the first answer is corrected according to the knowledge graph answer to obtain a second answer, and the second answer is used as the user answer.

7. The intelligent question-answering method for the electronic cigarette industry according to claim 1, characterized in that: Before optimizing the electronic cigarette industry big model according to the enriched and updated knowledge graph, the electronic cigarette industry intelligent question answering method further includes: A second question-answer pair is generated according to the user question and the user answer, and the knowledge graph is enriched and updated again using the second question-answer pair.

8. An intelligent question-answering device for the electronic cigarette industry, characterized in that: include: Knowledge graph construction module, used to collect and integrate e-cigarette industry knowledge and build a knowledge graph; An industry information module, used to train a large language model using the knowledge graph to obtain a large model of the e-cigarette industry; A knowledge graph enrichment and update module, used to enrich and update the knowledge graph using the electronic cigarette industry big model; A first answer generation module, used to obtain a first answer corresponding to the user's question according to the electronic cigarette industry macro model; A user answer generation module, used to compare and correct the first answer according to the knowledge graph to obtain a user answer; The model optimization module is used to optimize the electronic cigarette industry big model according to the enriched and updated knowledge graph.

9. The intelligent question-answering device for the electronic cigarette industry according to claim 8, characterized in that: Also includes: The knowledge graph is enriched and updated again by a module, which is used to generate a second question-answer pair according to the user question and the user answer, and to enrich and update the knowledge graph again by using the second question-answer pair.

10. An intelligent question-answering device for the electronic cigarette industry, characterized in that: The device comprises a processor and a memory: The memory is used to store program code and transmit the program code to the processor; The processor is used to execute the electronic cigarette industry intelligent question-answering method according to any one of claims 1-7 according to the instructions in the program code.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the electronic cigarette industry intelligent question-answering method according to any one of claims 1 to 7 is implemented.

Images