Method, device and equipment for automatically generating travel research report and storage medium

Abstract

The invention relates to the technical field of intelligent generation of research reports, and discloses an automatic generation method and device of a business research report, equipment and a storage medium. The method comprises the following steps: presetting an industry research report chapter catalogue; defining a standard work flow according to the industry research report chapter catalogue; and obtaining a research industry name input by a user, calling a corresponding search auxiliary tool and a large language model according to a standard workflow based on the research industry name, sequentially and automatically generating a content report of each chapter according to a chapter sequence, and obtaining a first version of an industry research report. According to the technical scheme, a user only needs to input the name of the target research industry, the industry research report which is complete in structure and detailed in content can be obtained in a short time, the problems that in traditional industry research report writing, consumed time is long, information is incomplete, and quality is inconsistent are effectively solved, and report generation efficiency and quality are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent generation of research reports, and in particular to a method, device, equipment and storage medium for automatic generation of research reports. Background Art

[0002] As market competition becomes increasingly fierce, the demand for industry research reports from enterprises, research institutions and investors has exploded. As an important basis for gaining insight into market dynamics, grasping industry trends and making strategic decisions, the quality and efficiency of industry research reports are crucial.

[0003] The traditional way of writing industry research reports mainly relies on manual operations, and professionals need to spend a lot of time and energy to collect data. This process involves obtaining relevant information from many different data sources, such as government databases, industry association reports, financial news websites, corporate annual reports, etc. According to statistics, it takes an average of 3-5 days to collect research data for a medium-sized industry, and professionals need to have a wide range of information channels and skilled search skills. After the data collection is completed, it is necessary to screen, organize and analyze the massive amount of data, which not only requires professionals to have deep industry knowledge and data analysis capabilities, but also is extremely time-consuming. The entire writing process usually takes 7-10 days or even longer. This method is inefficient and it is difficult to meet the market's requirements for the timeliness of industry research reports. In addition, the manual writing process is easily affected by subjective factors, resulting in problems such as data omissions and analysis bias. For example, when analyzing emerging industries, due to the rapid development of the industry and the complexity of information, some key information may be missed when manually screening data, which affects the accuracy and completeness of the report, and may mislead decision makers to make wrong judgments.

[0004] With the development of artificial intelligence technology, some report generation methods based on large language models have gradually emerged. For example, the invention patent with patent publication number CN118133973B discloses a report generation method and device based on a large language model, which mainly obtains the report key information input by the user, and uses the large language model to generate text questions corresponding to the report key information; converts it into a vector, and uses the vector to search in the vector database to obtain the corresponding approximate report, extracts the chapter directory of the approximate report as the directory to be processed; uses the directory to be processed to retrieve the content of the corresponding directory from the literature knowledge base, and assembles the directory to be processed and the corresponding content as the final report, thereby achieving the purpose of automatically generating a text report. Using the above patent method to generate a report, the user needs to input specific report key information in advance, and generate text corresponding to the key point based on the large language model, and then convert it into a vector to search in the vector database. For industry research reports with more complex content, this method relies on users to provide comprehensive and accurate key information, which increases the user's operating burden and may affect the quality of the report due to incomplete input of key points. In addition, the use of vector retrieval to extract similar chapter directories from existing reports may result in insufficient coherence in the generated report content if the directories do not match well or there are deviations, making it difficult to ensure comprehensive coverage of the target industry. Although this method speeds up report generation to a certain extent, in actual applications, it is still necessary to frequently adjust the directory and knowledge base content when targeting different industries and fields, which increases the workload of maintenance and adjustment and reduces the flexibility and applicability of the system.

[0005] Therefore, the prior art still needs to be improved and developed. Summary of the invention

[0006] In order to overcome the shortcomings of the prior art, the purpose of the present invention is to provide a method, device, equipment and storage medium for automatically generating industry research reports, aiming to solve the problem that the existing report generation method relies on users to provide comprehensive and accurate key information for industry research reports with more complex contents, which increases the user's operating burden and may affect the report quality due to incomplete input of key points.

[0007] A first aspect of the present invention provides a method for automatically generating industry research reports, comprising: pre-setting an industry research report catalog, wherein the industry research report catalog includes six chapters in sequence: industry definition and overview, macro analysis, meso analysis, supply chain analysis, micro analysis, summary and suggestions, and each chapter contains several sub-directories; defining a standard workflow according to the industry research report catalog; obtaining a research industry name input by a user, calling a corresponding search auxiliary tool and a large language model according to the standard workflow based on the research industry name, and automatically generating a content report for each chapter in sequence according to the chapter order to obtain a first edition of the industry research report.

[0008] Optionally, in a first implementation method of the first aspect of the present invention, the sub-directories included in the industry definition and summary are: industry definition, industry classification and characteristics; the sub-directories included in the macro-analysis are: macroeconomic environment, overall industry scale and output value, macro-policy impact; the sub-directories included in the meso-analysis are: technological innovation and trends, regional policy analysis, industry policies and regulations; the sub-directories included in the supply chain analysis are: industry supply chain structure, upstream and downstream relationships, supply chain management characteristics; the sub-directories included in the micro-analysis are: overview of major enterprises, comparison of enterprise competitiveness, market share analysis; the sub-directories included in the summary and suggestions are: summary of the current status of the industry, future development trends.

[0009] Optionally, in a second implementation of the first aspect of the present invention, a standard workflow is defined according to the industry research report chapter directory, including the steps of:

[0010] The standard workflow from Chapter 1 to Chapter 6 is defined according to the preset chapter order and sub-directory order in the chapter catalog of the industry research report. The standard workflow is used to guide the large language model to strictly follow the report structure to generate a content report.

[0011] Optionally, in a third implementation of the first aspect of the present invention, a research industry name input by a user is obtained, and a corresponding search auxiliary tool and a large language model are called according to a standard workflow based on the research industry name, and a content report for each chapter is automatically generated in sequence according to the chapter order to obtain a first edition of the industry research report, including the following steps: obtaining the sub-directory content of the current chapter, and calling a corresponding search auxiliary tool to perform a content search according to the research industry name input by the user and the sub-directory content to obtain a plurality of search results; inputting the plurality of search results into a large language model, the large language model including a content analysis module and a report generation module, the content analysis module being used to extract and summarize content information of the plurality of search results according to the sub-directory content to obtain a plurality of corresponding summary contents; the report generation module being used to embed the plurality of summary contents into a predefined prompt word template, and edit and integrate the summary contents according to the task requirements and task restrictions in the prompt word template to generate a content report related to the current sub-directory content; traversing all sub-directories of all chapters, calling corresponding search auxiliary tools and large language models, and automatically generating a content report for each sub-directory in sequence according to the chapter and sub-directory order to obtain a first edition of the industry research report.

[0012] Optionally, in a fourth implementation of the first aspect of the present invention, the search auxiliary tools include: a web search assistant, a listed company information query assistant, a document reading and parsing assistant, and a picture parsing and storage assistant.

[0013] Optionally, in a fifth implementation of the first aspect of the present invention, the large language model includes: one of GPT3.5, Moonshot-v1-32k and Moonshot-v1-128k.

[0014] Optionally, in the sixth implementation method of the first aspect of the present invention, after obtaining the first edition of the industry research report, it also includes the steps of editing, adding, deleting or regenerating specific chapters of the first edition of the industry research report displayed on the interactive interface as needed to meet personalized needs, and obtaining the final edition of the industry research report after confirmation and saving.

[0015] The second aspect of the present invention provides an automatic generation device for industry research reports, including: a chapter and section setting module, used to pre-set the chapter and section directory of the industry research report, the chapter and section directory of the industry research report includes six chapters in sequence: industry definition and overview, macro analysis, meso analysis, supply chain analysis, micro analysis, summary and suggestions, and each chapter contains several sub-directories; a definition module, used to define a standard workflow according to the chapter and section directory of the industry research report; a report generation module, used to obtain the research industry name input by the user, call the corresponding search auxiliary tool and large language model according to the standard workflow based on the research industry name, and automatically generate a content report for each chapter in sequence according to the chapter order to obtain a first edition of the industry research report.

[0016] The third aspect of the present invention provides an automatic generation device for industry research reports, comprising: a memory and at least one processor, wherein the memory stores computer-readable instructions, and the memory and the at least one processor are interconnected via lines; the at least one processor calls the computer-readable instructions in the memory to enable the automatic generation device for industry research reports to execute each step of the automatic generation method for industry research reports as described above.

[0017] A fourth aspect of the present invention provides a computer-readable storage medium, in which computer-readable instructions are stored. When the computer-readable storage medium is run on a computer, the computer executes each step of the method for automatically generating industry research reports as described above.

[0018] Beneficial effects: The present invention realizes the automatic generation of industry research reports through the combination of a large language model and a variety of search auxiliary tools. The solution first presets a standardized industry research report directory, including six main chapters such as industry definition and overview, macro analysis, meso analysis, and supply chain analysis; in this way, no matter what industry the report is, it can be written according to a fixed framework to ensure the logic and comprehensiveness of the content. During the report generation process, the user only needs to enter the target industry in the system, and the system will call the large language model according to the preset directory to generate the content of each chapter one by one. Through the technical solution of the present invention, the user only needs to enter the name of the target research industry to obtain a complete and detailed industry research report in a short time. This method effectively solves the problems of time-consuming, incomplete information, and inconsistent quality in the writing of traditional industry research reports, and significantly improves the efficiency and quality of report generation. At the same time, the scalability and flexibility of the system enable it to adapt to different industries and user needs, and has broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A flowchart of a method for automatically generating industry research reports provided in an embodiment of the present invention.

[0020] Figure 2 Generates a flow chart for the report contents of the subdirectories in the chapter.

[0021] Figure 3 This is a schematic diagram showing the initial version of the industry research report on the interactive interface.

[0022] Figure 4 This is a schematic diagram of the structure of the device for automatically generating industry research reports provided by the present invention.

[0023] Figure 5 This is a schematic diagram of the structure of the automatic generation equipment for industry research reports provided by the present invention. DETAILED DESCRIPTION

[0024] The embodiments of the present invention provide a method, device, equipment and storage medium for automatically generating industry research reports. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0025] The current process of writing industry research reports faces problems of inefficiency and information omissions. With the increasing demand for market insights from companies and investment institutions, especially in highly competitive fields such as finance, consumption and technology, the traditional way of writing industry research reports is time-consuming and laborious due to reliance on a large number of manual labor, and the accuracy of the report is easily affected by data omissions or analysis bias. In addition, existing report generation methods based on large language models also have certain limitations - for example, they usually require users to provide detailed report highlights, and in the case of complex content, the generation process may also have catalog mismatches or lack of coherence.

[0026] Based on this, the present invention provides a method for automatically generating a research report, such as Figure 1 As shown, it includes the steps of:

[0027] S10. Pre-set the industry research report chapter directory, which includes six chapters in sequence: industry definition and overview, macro analysis, meso analysis, supply chain analysis, micro analysis, summary and suggestions, and each chapter contains several sub-directories;

[0028] S20. Define a standard workflow according to the catalog of chapters in the industry research report;

[0029] S30. Obtain the name of the research industry input by the user, call the corresponding search auxiliary tool and large language model according to the standard workflow based on the research industry name, automatically generate the content report of each chapter in sequence according to the chapter order, and obtain the first edition of the industry research report.

[0030] Specifically, this embodiment pre-sets a standard industry research report catalog to make the structure of the industry research report clearer and more standardized, which is convenient for users to read and understand; at the same time, the unified directory structure is also conducive to the subsequent automatic generation of content reports according to the standard workflow, thereby improving the efficiency and quality of research report generation; the standard workflow is defined according to the industry research report catalog, and the operational steps and requirements of each link from data collection to content generation are clarified, making the entire research process more standardized and standardized, reducing interference from human factors, and improving the efficiency and quality of research report generation; at the same time, through the rational use of search assistance tools and large language models, data collection and analysis are automated, reducing labor costs.

[0031] Through the technical solution of this embodiment, users only need to enter the name of the target research industry to obtain a complete and detailed industry research report in a short time. This method effectively solves the problems of time-consuming, incomplete information, and inconsistent quality in the writing of traditional industry research reports, and significantly improves the efficiency and quality of report generation. At the same time, the scalability and flexibility of the system enable it to adapt to different industries and user needs, and has broad application prospects.

[0032] In some embodiments, the industry definition and summary include sub-categories such as industry definition, industry classification and characteristics; the macro-analysis includes sub-categories such as macroeconomic environment, overall industry scale and output value, and macro-policy impact; the meso-analysis includes sub-categories such as technological innovation and trends, regional policy analysis, and industry policies and regulations; the supply chain analysis includes sub-categories such as industry supply chain structure, upstream and downstream relationships, and supply chain management characteristics; the micro-analysis includes sub-categories such as overview of major companies, comparison of corporate competitiveness, and market share analysis; the summary and suggestions include sub-categories such as summary of the current status of the industry and future development trends.

[0033] Among them, by generating the content of the industry definition sub-directory, the connotation, boundaries and differences of the industry from other related industries can be clearly defined, which will help lay a solid foundation for subsequent research; it can enable readers to have an accurate understanding of the research object, avoid conceptual confusion during the research process, and ensure the pertinence and accuracy of the research. For example, when studying the new energy vehicle industry, clarifying its definition can distinguish different types such as pure electric vehicles and hybrid vehicles, so as to more accurately analyze the development of the industry. By generating the content of industry classification and characteristics sub-directories, researchers can examine the industry from different dimensions, discover the subdivisions and differences within the industry, and then tap into potential market opportunities and development trends; understanding the characteristics of the industry can highlight the uniqueness of the industry and provide a more targeted perspective for subsequent analysis; taking the clothing industry as an example, it can be subdivided into different fields such as fast fashion, high-end customization, and sportswear through classification. Each field has its own unique consumer groups, marketing strategies and competitive characteristics, which helps to gain a deep understanding of the overall picture of the industry.

[0034] By generating the content of the macroeconomic environment sub-directory, we can analyze the macroeconomic situation, such as the impact of GDP growth, inflation, interest rates, exchange rates and other factors on the industry, so that researchers can understand the general environment of the industry and grasp the law of the impact of macroeconomic fluctuations on the development of the industry; when the macroeconomic situation is good, some consumer industries often have development opportunities; and when the economic situation is not good, some basic industries may be hit harder; through the analysis of the macroeconomic environment, we can provide a macro-level basis for the prediction of industry development trends. Through the content of the overall scale and output value of the industry sub-directory, we can present the overall scale and output value data of the industry, which can intuitively reflect the development level of the industry and its position in the national economy. These data can help researchers and readers understand the development status of the industry, judge the development stage of the industry, and provide important reference for subsequent research and decision-making; for example, by comparing the industry output value data of different years, we can judge the growth rate and development trend of the industry. Through the content of the macro-policy impact sub-directory, we can collect and interpret the policies and regulations related to the industry issued by the national and local governments, evaluate the impact of policies on the development of the industry, and enable researchers to timely understand policy directions and grasp policy opportunities and challenges; policies have an important guiding and regulatory role in the development of the industry. Some support policies may promote the rapid development of the industry, while some restrictive policies may have certain constraints on the development of the industry; through the analysis of the impact of macro-policies, it can provide decision-making references for enterprises and investors, helping them better adapt to changes in the policy environment.

[0035] By generating the content of the sub-directory of technological innovation and trends, we can explore the technological innovation trends related to the industry, such as the application of new technologies, R&D investment, etc., which promote the development of the industry. This will enable researchers to understand the technological development dynamics of the industry and grasp the opportunities and challenges brought by technological innovation. Through the analysis of technological innovation and trends, we can help enterprises and investors to make early arrangements and seize the commanding heights of technological innovation. By generating the content of the sub-directory of regional policy analysis, we can analyze the impact of policy differences in different regions on the development of the industry, and enable researchers to understand the development environment and policy support of the industry in different regions, and provide reference for the regional layout and investment decisions of enterprises. Different policy orientations and resource endowments in different regions have different impacts on the development of the industry. For example, some regions may introduce preferential policies to attract high-tech enterprises to settle in, while other regions may pay more attention to the transformation and upgrading of traditional industries; through regional policy analysis, we can help enterprises choose regions that are more suitable for their own development. By generating the content of the sub-directory of industry policies and regulations, researchers can understand the norms and standards of the industry and provide guidance for the compliance operation of enterprises. Policies and regulations play an important role in regulating and constraining the development of industries. The entry barriers and product standards of some industries are affected by policies and regulations. By analyzing industry policies and regulations, companies can understand the development rules of the industry and avoid risks caused by violations.

[0036] By generating sub-directory contents of the industry supply chain structure, researchers can understand the upstream and downstream relationships and industrial layout of the industry, and provide a basis for analyzing the stability and competitiveness of the supply chain; by analyzing the supply chain structure, weak links and potential risks in the supply chain can be discovered, providing a reference for the supply chain management of enterprises; for example, in the electronic products industry, the stability of the supply chain is crucial to the production and sales of enterprises. By analyzing the supply chain structure, enterprises can optimize the supply chain layout and improve the stability and efficiency of the supply chain. By generating sub-directory contents of upstream and downstream relationships, the relationship between upstream and downstream enterprises can be analyzed, such as the bargaining power of suppliers and the characteristics of customer demand, which can enable researchers to understand the operation mechanism and value transfer process of the supply chain and provide guidance for the supply chain management of enterprises; the relationship between upstream and downstream enterprises has an important impact on the cost control, product quality and market competitiveness of enterprises; by analyzing the upstream and downstream relationships, enterprises can optimize supply chain management and improve their competitiveness. By generating sub-directory content of supply chain management characteristics, we can study the characteristics of supply chain management, such as inventory management, logistics distribution, etc., which can enable researchers to understand the industry's supply chain management model and efficiency and provide reference for the company's supply chain management; different industries have different supply chain management characteristics, some industries may pay more attention to inventory management, while other industries may pay more attention to logistics distribution efficiency; through the analysis of supply chain management characteristics, it can help companies learn advanced supply chain management experience and improve their own supply chain management level.

[0037] By generating the sub-directory content of the main enterprise profile, researchers can understand the main players in the industry, providing a basis for subsequent enterprise competitiveness comparison and market share analysis; by analyzing the main enterprise profiles, the advantages and disadvantages of the enterprise can be discovered, providing a reference for the development of the enterprise. By generating the sub-directory content of enterprise competitiveness comparison, the competitiveness of major enterprises can be compared, such as the technical strength, brand influence, marketing capabilities, etc. of the enterprise, which can enable researchers to understand the competitive landscape and competitive position of the enterprise in the industry, and provide a reference for the strategic decision-making of the enterprise; by comparing the competitiveness of enterprises, it can help enterprises discover their own advantages and disadvantages and formulate corresponding competitive strategies. By generating the sub-directory content of market share analysis, researchers can understand the degree of market competition in the industry and the market position of the enterprise, and provide a reference for the market expansion and investment decision-making of the enterprise; market share is one of the important indicators for measuring the competitiveness of an enterprise. By analyzing the market share, it can help enterprises understand the market demand and competitive situation and formulate reasonable market strategies.

[0038] By generating a sub-directory content of the summary of the current status of the industry, readers can quickly understand the current status of the industry and research results, and provide reference for subsequent decision-making; the summary of the current status of the industry is a summary and refinement of the entire research report, which can help readers grasp the key points and improve reading efficiency. By generating a sub-directory content of future development trends, the future development trends of the industry can be predicted, including market size, competition landscape, technological innovation, etc., which can provide forward-looking guidance for enterprises and investors, helping them to plan ahead and seize market opportunities; the prediction of future development trends is one of the important contents of industry research reports, which can help enterprises and investors understand the development direction of the industry and formulate reasonable development strategies.

[0039] This embodiment can ensure the structural integrity and logic of the report by presetting a unified chapter directory and chapter subdirectory, and provide a clear framework for subsequent content generation.

[0040] In some embodiments, a standard workflow is defined according to the industry research report catalog, including the steps of: defining a standard workflow from Chapter 1 to Chapter 6 according to a preset chapter order and sub-directory order in the industry research report catalog, wherein the standard workflow is used to guide a large language model to strictly follow the report structure to generate a content report.

[0041] Specifically, by defining a standard workflow, the large language model can be guided to generate content strictly according to the process. The chapters and sub-directories are logically coherent, and the analysis is gradually deepened from macro to micro, making the report structure rigorous and easy for readers to understand. For example, after the macro analysis, the meso analysis can be naturally transitioned, allowing readers to switch from the perspective of the overall economic environment to the perspective of the internal structure of the industry. At the same time, a clear standard workflow can also ensure that each chapter and sub-directory has corresponding data collection and analysis steps to avoid content omissions. For example, in the supply chain analysis, the process from structure to upstream and downstream relationships to management characteristics ensures that all aspects of the supply chain are fully covered. The standard workflow allows the large language model to generate content in a regular manner, reduce confusion and repetitions in the generation process, and increase the speed of report generation.

[0042] In some implementations, obtaining a research industry name input by a user, calling a corresponding search auxiliary tool and a large language model according to a standard workflow based on the research industry name, and automatically generating a content report for each chapter in sequence according to the chapter order to obtain a preliminary version of the industry research report includes the following steps:

[0043] Get the sub-directory content of the current chapter, call the corresponding search auxiliary tool to search the content according to the research industry name and sub-directory content entered by the user, and obtain multiple search results;

[0044] Inputting the plurality of search results into a large language model, the large language model comprising a content analysis module and a report generation module, the content analysis module being used to extract and summarize content information of the plurality of search results according to the sub-directory content to obtain a plurality of corresponding summary contents; the report generation module being used to embed the plurality of summary contents into a predefined prompt word template, and editing and integrating the summary contents according to the task requirements and task restrictions in the prompt word template to generate a content report related to the current sub-directory content;

[0045] Traverse all subdirectories of all chapters, call corresponding search assistance tools and large language models, and automatically generate content reports for each subdirectory in sequence according to the order of chapters and subdirectories to obtain the first edition of the industry research report.

[0046] Specifically, if Figure 2 As shown, for each subdirectory in the six chapters in this embodiment, the content report generation process needs to be repeated. Figure 2 As an example, after obtaining the name of the research industry (such as the new energy industry) input by the user, the content report of the sub-directory (industry definition) in Chapter 1 (industry definition and summary) is first generated according to the standard workflow; in the process of generating the content report, first Figure 2The search module in the system determines the key search terms (e.g., “new energy industry definition”) based on the research industry name (new energy industry) and sub-directory content (industry definition) input by the user, and then calls the corresponding search auxiliary tool to search the content based on the key search terms, and obtains the following results: Figure 2 The multiple search results shown are then input into the analysis module in the large language model, and the analysis module will extract and summarize the content information of the multiple search results according to the sub-directory content (industry definition), and obtain multiple summary contents related to the new energy industry definition; then the multiple summary contents are embedded into the predefined prompt word template, and the multiple summary contents are edited and integrated in the generation module according to the task requirements and task restrictions in the prompt word template to generate a content report related to the current sub-directory content, thus completing the content report of the industry definition of the first sub-directory of Chapter 1. According to the above steps, the content report of each sub-directory of each chapter is generated in turn to obtain the first version of the industry research report.

[0047] In this embodiment, the generation module is preset with a prompt word template, which is a templated tool for asking questions to the large language model. It provides some "hints" to the large language model through a fixed structure to help it better understand the task, thereby improving the performance of the model; the content framework of the prompt word template in this embodiment includes: task requirements, reference information (i.e., the summary content generated by the analysis module) and task restrictions; for different sub-directories, the task requirements, reference information and task restrictions are different. As an example, the task requirement refers to the task that the large language model needs to complete in the current dialogue. If you want to generate a content report on the industry definition of the first sub-directory of Chapter 1, the task requirement at this time is: write a new energy industry definition for the reference information; the task restriction is usually fixed in the system, and the restrictions for the generation of different sub-sections are slightly different. For example, for the industry definition part of Chapter 1, the restriction usually requires the large language model to write a text of no less than 200 words, and for the micro-analysis of the enterprise profile part, the restriction usually requires the large language model to write a text of no less than 600 words.

[0048] In this embodiment, the search assistance tool includes: a web search assistant, a listed company information query assistant, a document reading and parsing assistant, and a picture parsing and storage assistant. Among them, the web search assistant is used to retrieve the latest industry information, news, policies and data on the public network in real time; the assistant is implemented based on the API of the searXNG search engine. By calling the API and then entering the industry name to search on the public network, the industry definition, market dynamics, policy documents, etc. can be obtained. The listed company information query assistant can be implemented by calling the API of the two websites of Yahoo Finance and Eastmoney. After entering the company name, the company information can be queried to obtain the financial statements, operating conditions, market share and other information of the relevant company; the listed company information query assistant can provide detailed data support for major companies in the industry. The document reading and parsing assistant can be implemented by calling the API of the jina.ai document reading tool. After the local document is passed in, all the contents in it can be read and parsed; it also supports users to upload documents and extract the information therein to provide to the large language model as a knowledge reference. The image parsing and storage assistant is implemented by calling Baidu's image recognition API. After the image is passed in, relevant information can be obtained from it and provided to the large language model as a knowledge reference; the assistant uses Baidu's image recognition / image classification API to support image recognition and classification; at the same time, the large language model determines that the relevant images are useful and will be synchronously nested and displayed in the content of the corresponding chapter, and the user can choose to keep or delete them.

[0049] In this embodiment, for the generation of content reports for different chapters, different search assistance tools can be selected to collect data. As an example, for the industry definition and overview chapters, you can call a web search assistant to obtain the definition, basic overview and development background of the target industry, such as Baidu Encyclopedia, industry association websites, etc.; for the macro analysis chapter, you can use a web search assistant and a document reading and parsing assistant to collect macroeconomic environment data, the overall scale of the industry, market output value, macro policies and other information, and obtain data from sources such as the National Bureau of Statistics, central bank reports, and government policy documents; for the meso analysis chapter, you can call a web search assistant and a document reading and parsing assistant to obtain the industry's technological innovation trends, regional policies, industry regulations and other information, and collect Ministry of Science and Technology reports, local government policy documents, industry standards and other information. and other materials; for the supply chain analysis chapter, you can use the web search assistant, document reading and parsing assistant, and image parsing and storage assistant to collect information on the industry's supply chain structure, upstream and downstream companies, suppliers and distributors, and obtain reports from the Supply Chain Management Association, industry white papers, etc.; for the micro-analysis chapter, you can use the listed company information query assistant to obtain the financial data, market share, competitive advantages, etc. of major companies in the industry, and query stock exchanges, company annual reports, investor relations websites, etc.; for the summary and suggestions chapter, you can call the web search assistant to obtain the latest market forecasts and expert opinions, and make a comprehensive summary based on the content of the first five chapters and the latest industry information.

[0050] In this embodiment, the large language model can select any one of GPT3.5, Moonshot-v1-32k and Moonshot-v1-128k, but is not limited thereto. The large language model is responsible for generating natural language text and writing various parts of the report based on the input information and the data provided by the search auxiliary tool. It has the ability to understand the industry background, analyze data and generate fluent text. Among them, GPT3.5 is a generative artificial intelligence language dialogue model developed by OpenAI; Moonshot-v1 is a generative artificial intelligence language dialogue model developed by China's Moonshot technology startup. 32k and 128k indicate that the maximum context length supported by the language model is 32,000 characters and 128,000 characters respectively.

[0051] In some embodiments, after obtaining the first edition of the industry research report, the step is also included: editing, adding, deleting or regenerating specific chapters of the first edition of the industry research report displayed on the interactive interface as needed to meet personalized needs, and obtaining the final edition of the industry research report after confirmation and saving.

[0052] Specifically, if Figure 3As shown in the figure, the first version of the industry research report is displayed on the interactive interface, and users can edit, add, delete or regenerate specific chapters of the first version of the industry research report. Specific functions include: content editing, that is, users can modify text, adjust wording or add new information directly on the interface; they can also delete or add image links by themselves; regenerate specific chapters, that is, for unsatisfactory content (usually a subdirectory), users can request regeneration, at which time the large language model and search assistance tools will be re-called to generate new content for users to review; the final confirmed report can save the current version or export it to PDF format.

[0053] The technical solution of this embodiment has the following advantages: Standardized report chapter catalog: 6 main chapters are preset to ensure the logic and structure of the report, which is suitable for the reporting needs of most industries; Automated data collection and integration: Integrated search auxiliary tools such as web search assistant, listed company information query assistant, document parsing assistant, etc., automatically obtain industry data from multiple sources to ensure the richness and timeliness of information; Multi-channel data support combined with large language model: public data is obtained through auxiliary tools and provided to the large language model to achieve accurate and detailed content generation; Efficient generation process: define the standard workflow from Chapter 1 to Chapter 6, and automatically generate in the order of chapters to ensure the orderliness of content generation and reduce human intervention; Intelligent content generation and improvement mechanism: users can choose to regenerate specific chapters or update data according to their needs to further enhance the personalization and accuracy of the content.

[0054] Through the technical solution of this embodiment, users only need to enter the name of the target industry to obtain a complete and detailed industry research report in a short time. This method effectively solves the problems of time-consuming, incomplete information, and inconsistent quality in the writing of traditional industry research reports, and significantly improves the efficiency and quality of report generation. At the same time, the scalability and flexibility of the system enable it to adapt to different industries and user needs, and has broad application prospects.

[0055] The above describes the automatic generation method of the research report in the embodiment of the present invention. The following describes the automatic generation device of the research report in the embodiment of the present invention. Figure 4 An embodiment of the automatic generation device of the research report in the embodiment of the present invention includes:

[0056] The chapter directory setting module 10 is used to pre-set the chapter directory of the industry research report, and the chapter directory of the industry research report includes six chapters in sequence: industry definition and overview, macro analysis, meso analysis, supply chain analysis, micro analysis, summary and suggestions, and each chapter contains a number of sub-directories;

[0057] A definition module 20, used to define a standard workflow according to the industry research report chapter directory;

[0058] The report generation module 30 is used to obtain the research industry name input by the user, call the corresponding search auxiliary tools and large language model according to the standard workflow based on the research industry name, automatically generate the content report of each chapter in sequence according to the chapter order, and obtain the first edition of the industry research report.

[0059] Based on the same idea as the method in the above-mentioned embodiment, the device provided in the present application can implement the method in the above-mentioned embodiment. For the convenience of explanation, the structural diagram of the device embodiment only shows the parts related to the embodiment of the present application. Those skilled in the art can understand that the illustrated structure does not constitute a limitation on the device, and may include more or fewer modules than shown in the figure, or a combination of certain modules, or different module arrangements.

[0060] Figure 4 The automatic generation device of industry research reports in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The automatic generation device of industry research reports in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0061] Figure 5 It is a structural diagram of an automatic generation device for industry research reports provided by an embodiment of the present invention. The automatic generation device 100 for industry research reports may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 11 (for example, one or more processors) and a memory 12, and one or more storage media 13 (for example, one or more massive storage devices) storing application programs 133 or data 132. Among them, the memory 12 and the storage medium 13 may be short-term storage or permanent storage. The program stored in the storage medium 13 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the automatic generation device 100 for industry research reports. Furthermore, the processor 11 may be configured to communicate with the storage medium 13 to execute a series of instruction operations in the storage medium 13 on the automatic generation device 100 for industry research reports.

[0062] The automatic generation device 100 for industry research reports may also include one or more power supplies 14, one or more wired or wireless network interfaces 15, one or more input and output interfaces 16, and / or one or more operating systems 131, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, etc. It will be appreciated by those skilled in the art that Figure 5The device structure shown does not constitute a limitation on the automatic generation device for industry research reports 100, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0063] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the method for automatically generating a research report.

[0064] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, device, or unit can refer to the corresponding process in the aforementioned method embodiment and will not be repeated here.

[0065] If the integrated unit is implemented in the form of a software functional unit 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 invention 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. The computer software product is stored in a storage medium, including several instructions for 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 method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0066] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for automatically generating a research report, characterized in that: Includes steps: Pre-set the industry research report chapter directory, which includes six chapters in sequence: industry definition and overview, macro analysis, meso analysis, supply chain analysis, micro analysis, summary and suggestions, and each chapter contains several sub-directories; Define standard workflows based on the industry research report chapter catalog; Obtain the research industry name input by the user, call the corresponding search auxiliary tool and large language model according to the standard workflow based on the research industry name, automatically generate the content report of each chapter in sequence according to the chapter order, and obtain the first edition of the industry research report.

2. The method for automatically generating industry research reports according to claim 1, characterized in that: The sub-categories of the industry definition and summary are: industry definition, industry classification and characteristics; the sub-categories of the macro analysis are: macroeconomic environment, overall industry scale and output value, macro policy impact; the sub-categories of the meso analysis are: technological innovation and trends, regional policy analysis, industry policies and regulations; The sub-categories of the supply chain analysis include: industry supply chain structure, upstream and downstream relationships, and supply chain management characteristics; the sub-categories of the micro-analysis include: overview of major companies, comparison of corporate competitiveness, and market share analysis; the sub-categories of the summary and suggestions include: summary of the current industry situation and future development trends.

3. The method for automatically generating industry research reports according to claim 2, characterized in that: Define a standard workflow based on the industry research report chapter directory, including the following steps: The standard workflow from Chapter 1 to Chapter 6 is defined according to the preset chapter order and sub-directory order in the chapter catalog of the industry research report. The standard workflow is used to guide the large language model to strictly follow the report structure to generate a content report.

4. The method for automatically generating a research report according to claim 3, characterized in that: Obtain the research industry name input by the user, call the corresponding search auxiliary tool and large language model according to the standard workflow based on the research industry name, automatically generate the content report of each chapter in sequence according to the chapter order, and obtain the first edition of the industry research report, including the following steps: Get the sub-directory content of the current chapter, call the corresponding search auxiliary tool to search the content according to the research industry name and sub-directory content entered by the user, and obtain multiple search results; Inputting the plurality of search results into a large language model, the large language model comprising a content analysis module and a report generation module, the content analysis module being used to extract and summarize content information of the plurality of search results according to the sub-directory content to obtain a plurality of corresponding summary contents; the report generation module being used to embed the plurality of summary contents into a predefined prompt word template, and editing and integrating the summary contents according to the task requirements and task restrictions in the prompt word template to generate a content report related to the current sub-directory content; Traverse all subdirectories of all chapters, call corresponding search assistance tools and large language models, and automatically generate content reports for each subdirectory in sequence according to the order of chapters and subdirectories to obtain the first edition of the industry research report.

5. The method for automatically generating industry research reports according to claim 4, characterized in that: The search auxiliary tools include: a web page search assistant, a listed company information query assistant, a document reading and parsing assistant, and a picture parsing and storage assistant.

6. The method for automatically generating industry research reports according to claim 4, characterized in that: The large language model includes: one of GPT3.5, Moonshot-v1-32k and Moonshot-v1-128k.

7. The method for automatically generating industry research reports according to claim 1, characterized in that: After obtaining the initial version of the industry research report, the following steps are also included: According to needs, the initial version of the industry research report displayed on the interactive interface can be edited, added, deleted or specific chapters can be regenerated to meet personalized needs. After confirmation and saving, the final version of the industry research report can be obtained.

8. A device for automatically generating industry research reports, characterized in that: include: A chapter directory setting module is used to pre-set the chapter directory of the industry research report. The chapter directory of the industry research report includes six chapters in sequence: industry definition and overview, macro analysis, meso analysis, supply chain analysis, micro analysis, summary and suggestions. Each chapter contains several sub-directories. A definition module, used to define a standard workflow according to the chapter directory of the industry research report; The report generation module is used to obtain the research industry name input by the user, call the corresponding search auxiliary tools and large language model according to the standard workflow based on the research industry name, automatically generate the content report of each chapter in sequence according to the chapter order, and obtain the first edition of the industry research report.

9. An automatic generation device for industry research reports, characterized in that: comprising a memory and at least one processor, wherein the memory has computer-readable instructions stored therein; The at least one processor calls the computer-readable instructions in the memory to execute the various steps of the method for automatically generating industry research reports as described in any one of claims 1-7.

10. A computer-readable storage medium having computer-readable instructions stored thereon, characterized in that: When the computer-readable instructions are executed by a processor, the various steps of the method for automatically generating industry research reports as described in any one of claims 1 to 7 are implemented.

Images