A professional report generation system, method, electronic device and storage medium

Abstract

The application discloses a professional report generation system and method, an electronic device and a storage medium. The application analyzes and judges whether auxiliary information is needed according to a report outline generated by a large language model based on a guiding title and a report outline input by a target object; generates a calling instruction when the auxiliary information is needed; searches for the auxiliary information in response to the calling instruction through a search enhancement subsystem; generates a report outline by calling the large language model based on the guiding title and knowledge data through a report outline generation subsystem; and generates a report body by calling the large language model based on the report outline and the knowledge data through a report body generation subsystem. The application can accurately describe the report content, supports quick customization to meet the needs of specific industries, and provides a more intelligent, efficient and customized professional report generation scheme, significantly improves the work efficiency and quality of users in writing complex and important documents, and can be widely applied to the technical field of data processing.

Description

[0001] This invention patent application is a divisional application of the invention patent application with application number 202510393691.4, application date March 31, 2025, and invention title "A professional report generation system, method, electronic device and storage medium". Technical Field

[0002] This invention relates to the field of data processing technology, and in particular to a professional report generation system, method, electronic device, and storage medium. Background Technology

[0003] With the rapid development of artificial intelligence and machine learning technologies, Large Language Models (LLMs) have made significant progress in the field of Natural Language Processing (NLP). These models, trained on large-scale datasets, are capable of understanding and generating human language and are widely used in text generation, dialogue systems, translation services, and many other areas. Currently, in the generation of professional reports, existing technologies mainly include template filling, rule-based methods, and early language model-assisted writing. Template filling relies on predefined structures and content frameworks; users only need to fill in specific information. Rule-based methods rely on a series of pre-defined logical and grammatical rules to construct sentences and paragraphs. While early language models can generate text based on given context, their understanding and generation capabilities are relatively limited.

[0004] Current technologies for generating professional reports mainly rely on template filling, rule-based methods, and early language models. While these methods have achieved automated text generation to some extent, they suffer from insufficient flexibility and struggle to adapt to the complex and ever-changing needs of professional fields, particularly when in-depth analysis of "current situation, problems, and solutions" is required. Furthermore, early language models have limited ability to handle complex logical relationships and provide in-depth insights, failing to meet the requirements of high-quality professional reports. They are also difficult to customize and cannot be adjusted to the specific requirements of different industries or fields. Summary of the Invention

[0005] This invention proposes a professional report generation system, method, electronic device, and storage medium, aiming to at least partially solve one of the technical problems in related technologies. The embodiments of this invention can efficiently and accurately generate professional reports.

[0006] On one hand, embodiments of the present invention provide a professional report generation system, the system including a large language model layer and a business logic layer, the business logic layer including a retrieval enhancement subsystem, a report outline generation subsystem and a report body generation subsystem; The large language model layer has a pre-set large language model; the large language model is used to analyze and determine whether auxiliary information is needed based on the guiding title and report outline generated by the subsystem based on the input of the target object; when auxiliary information is needed, it generates a calling instruction; The retrieval enhancement subsystem is used to retrieve auxiliary information in response to invocation commands; The report outline generation subsystem is used to generate a report outline based on guiding titles and knowledge data, by calling a large language model; the knowledge data includes data retrieved from the internal knowledge base; when auxiliary information exists, the knowledge data also includes auxiliary information. The report text generation subsystem is used to generate the report text based on the report outline and knowledge data by calling a large language model.

[0007] Optionally, the auxiliary information includes first auxiliary information and second auxiliary information; the retrieval data includes first retrieval data and second retrieval data; when the large language model analyzes and determines whether auxiliary information is needed based on the guiding title input by the target object and the report outline generated by the report outline generation subsystem, it is specifically used to perform the following operations: Semantic analysis of the guiding titles reveals the subject areas; The first search data is obtained from the internal knowledge base based on the subject area; If the amount of data in the first search meets the preset conditions, it is determined that the first auxiliary information is not needed; otherwise, it is determined that the first auxiliary information is needed. Based on the report outline, text prompts are constructed, and second-level search data is obtained from the internal knowledge base through the text prompts; If the amount of data in the second retrieval meets the preset conditions, it is determined that the second auxiliary information is not needed; otherwise, it is determined that the second auxiliary information is needed.

[0008] Optionally, the retrieval enhancement subsystem includes a local knowledge base retrieval unit and an internet retrieval unit; when the retrieval enhancement subsystem retrieves auxiliary information in response to a call command, it specifically performs the following operations: In response to the call command, based on the search information, the system retrieves structured information from the pre-built local knowledge base through the local knowledge base retrieval unit, and obtains external information resources through the Internet retrieval unit. Auxiliary information is obtained by integrating structured information and external information resources; The invocation instruction includes a first invocation instruction and a second invocation instruction; when the invocation instruction is the first invocation instruction, the retrieved information adopts the subject domain, and the auxiliary information is the first auxiliary information; when the invocation instruction is the second invocation instruction, the retrieved information adopts text prompts, and the auxiliary information is the second auxiliary information.

[0009] Optionally, the local knowledge base includes a data source system, a retrieval system, and intelligent summarization; the retrieval enhancement subsystem also includes a local knowledge base construction unit, which is specifically used to perform the following operations: The pre-collected data sources are standardized to obtain a data source system; the standardization process includes format unification conversion, text structure parsing, and key information point extraction; the data sources are dynamically updated periodically based on a preset cycle; Based on the processing results of key information point extraction, a retrieval system is constructed using a multi-level knowledge tree structure; the multi-level knowledge tree structure includes subject areas, sub-topics, and specific cases / policies; Based on the retrieval system, a large language model is invoked to generate intelligent summaries step by step from specific cases / policies.

[0010] Optionally, the report outline generation subsystem includes a current situation summary generation unit, a problem and challenge analysis unit, and a countermeasures and suggestions generation unit; the report outline includes a current situation summary, problem and challenge analysis data, and countermeasures and suggestions; when generating the report outline based on guiding titles and knowledge data, and calling a large language model, the report outline generation subsystem is specifically used to perform the following operations: The current situation summary generation unit uses the guiding title and knowledge data to call the large language model to accurately extract the current situation summary. Based on the current situation, the problem and challenge analysis unit summarizes the current situation and uses a large language model for analysis and reasoning to obtain problem and challenge analysis data. The countermeasures and suggestions generation unit analyzes the current situation and challenges based on the data, and then uses a large language model to perform suggestion analysis to obtain countermeasures and suggestions.

[0011] Optionally, the report text generation subsystem includes a report text generation unit; when generating the report text based on the report outline and knowledge data by calling the large language model, the report text generation subsystem specifically performs the following operations: The report outline is broken down and transformed to obtain the guiding / topic sentence corresponding to each outline item; Text prompts are generated based on the guiding / topic sentences; Based on text prompts, the system retrieves supporting data from knowledge data, and then calls a large language model to integrate the content and generate the report text corresponding to each outline item.

[0012] Optionally, the report body generation subsystem also includes a report summary generation unit, which is specifically used to perform the following operations: Based on all the report texts, the core viewpoints and key conclusions are extracted by using a large language model, and then combined with the report texts to construct prompt texts. Based on the prompt text, the large language model is invoked to generate a report summary.

[0013] On the other hand, embodiments of the present invention provide a professional report generation method, applied to the aforementioned professional report generation system, the method comprising the following steps: The large language model analyzes the report outline generated by the subsystem based on the guiding title and report outline input by the target object to determine whether auxiliary information is needed; if auxiliary information is needed, it generates a calling instruction. The retrieval enhancement subsystem retrieves auxiliary information in response to the call command. The report outline generation subsystem generates a report outline based on guiding titles and knowledge data, and calls a large language model to generate the outline. The knowledge data includes data retrieved from the internal knowledge base; when auxiliary information is available, the knowledge data also includes auxiliary information. The report text generation subsystem generates the report text based on the report outline and knowledge data by calling a large language model.

[0014] Optionally, the local knowledge base includes a data source system, a retrieval system, and intelligent summarization; the method also includes the following steps: The pre-collected data sources are standardized to obtain a data source system; the standardization process includes format unification conversion, text structure parsing, and key information point extraction; the data sources are dynamically updated periodically based on a preset cycle; Based on the processing results of key information point extraction, a retrieval system is constructed using a multi-level knowledge tree structure; the multi-level knowledge tree structure includes subject areas, sub-topics, and specific cases / policies; Based on the retrieval system, a large language model is invoked to generate intelligent summaries step by step from specific cases / policies.

[0015] Optionally, the method further includes the following steps: Based on all the report texts, the core viewpoints and key conclusions are extracted by using a large language model, and then combined with the report texts to construct prompt texts. Based on the prompt text, the large language model is invoked to generate a report summary.

[0016] On the other hand, embodiments of the present invention provide an electronic device, including: a processor and a memory; the memory is used to store a program; the processor executes the program to implement the above-mentioned professional report generation method.

[0017] On the other hand, embodiments of the present invention provide a computer storage medium storing a processor-executable program, which, when executed by a processor, is used to implement the aforementioned professional report generation method.

[0018] This invention, through a large language model, analyzes the report outline generated by the target object's input guiding title and report outline generation subsystem to determine whether auxiliary information is needed. If auxiliary information is needed, a call command is generated. The retrieval enhancement subsystem retrieves the auxiliary information in response to the call command. The report outline generation subsystem, based on the guiding title and knowledge data, calls the large language model to generate the report outline. The knowledge data includes retrieval data from an internal knowledge base; if auxiliary information exists, the knowledge data also includes the auxiliary information itself. The report text generation subsystem, based on the report outline and knowledge data, calls the large language model to generate the report text. This invention offers the following advantages: By utilizing the highly flexible understanding and generation capabilities of an advanced large language model, it can accurately describe report content while supporting rapid customization to meet the needs of specific industries. This provides a more intelligent, efficient, and customized professional report generation solution, significantly improving the efficiency and quality of users writing complex and important documents. Attached Figure Description

[0019] The accompanying drawings are provided to further understand the technical solutions of the present invention and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the technical solutions of the present invention, and do not constitute a limitation on the technical solutions of the present invention.

[0020] Figure 1 A schematic diagram illustrating the module composition of the professional report generation system provided in this embodiment of the invention; Figure 2 A schematic diagram illustrating the construction process of a local knowledge base provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the overall architecture of the professional report generation system provided in an embodiment of the present invention; Figure 4 A schematic diagram of an implementation environment for a professional report generation method provided in an embodiment of the present invention; Figure 5 A flowchart illustrating a professional report generation method provided in an embodiment of the present invention; Figure 6 A schematic diagram illustrating the overall process of the professional report generation method provided in this embodiment of the invention; Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0022] It should be noted that although functional modules are divided in the system diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the system or the order in the flowchart. The terms "first / S100," "second / S200," etc., in the specification, claims, and the aforementioned figures are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0023] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0024] On the one hand, such as Figure 1 As shown, this embodiment of the invention provides a professional report generation system 100, which may include a large language model layer 110 and a business logic layer 120. The business logic layer includes a retrieval enhancement subsystem 121, a report outline generation subsystem 122, and a report body generation subsystem 123. The large language model layer has a pre-set large language model; the large language model is used to analyze and determine whether auxiliary information is needed based on the guiding title and report outline generated by the subsystem based on the input of the target object; when auxiliary information is needed, it generates a calling instruction; For example, in some specific implementations, the large language model layer serves as the foundation of the entire system. This layer fully leverages the superior text generation and understanding capabilities of the large language model, providing solid technical support for the "report outline generation subsystem" and "report body generation subsystem" in the business logic layer. By deeply integrating the powerful functions of the large language model, not only can the accuracy and rationality of these two subsystems in constructing the report structure be ensured, but their richness and professionalism in content creation can also be significantly enhanced. As the intelligent core, the large language model enables the system to demonstrate outstanding capabilities in parsing complex information, capturing subtle semantics, and generating coherent and insightful text, thereby laying the foundation for ultimately generating high-quality professional reports.

[0025] It should be noted that in some embodiments, auxiliary information includes first auxiliary information and second auxiliary information; retrieval data includes first retrieval data and second retrieval data; when the large language model analyzes and determines whether auxiliary information is needed based on the guiding title input by the target object and the report outline generated by the report outline generation subsystem, it is specifically used to perform the following operations: Semantic analysis of the guiding titles reveals the subject areas; The first search data is obtained from the internal knowledge base based on the subject area; If the amount of data in the first search meets the preset conditions, it is determined that the first auxiliary information is not needed; otherwise, it is determined that the first auxiliary information is needed. Based on the report outline, text prompts are constructed, and second-level search data is obtained from the internal knowledge base through the text prompts; If the amount of data in the second retrieval meets the preset conditions, it is determined that the second auxiliary information is not needed; otherwise, it is determined that the second auxiliary information is needed.

[0026] For example, in some implementations, after receiving a title, the large language model performs semantic analysis to determine the subject area of ​​the report and assess whether additional information is needed to ensure the accuracy and depth of the content. This includes checking whether there is sufficient relevant data in the internal knowledge base or whether there is information that needs to be updated.

[0027] In addition, a second assessment of the need for supporting information can be conducted: Before writing the main text, to ensure that the large model can fully utilize more comprehensive and accurate supporting information during the generation process, the model will conduct a second assessment based on the generated report outline to determine whether additional supporting information is needed to support detailed arguments. This process aims to enhance the depth and accuracy of the content, ensuring that each paragraph is fully supported and argued.

[0028] The retrieval enhancement subsystem is used to retrieve auxiliary information in response to invocation commands; It should be noted that, in some embodiments, the retrieval enhancement subsystem includes a local knowledge base retrieval unit and an internet retrieval unit; when the retrieval enhancement subsystem retrieves auxiliary information in response to a call command, it specifically performs the following operations: In response to the call command, based on the search information, the system retrieves structured information from the pre-built local knowledge base through the local knowledge base retrieval unit, and obtains external information resources through the Internet retrieval unit. Auxiliary information is obtained by integrating structured information and external information resources; The invocation instruction includes a first invocation instruction and a second invocation instruction; when the invocation instruction is the first invocation instruction, the retrieved information adopts the subject domain, and the auxiliary information is the first auxiliary information; when the invocation instruction is the second invocation instruction, the retrieved information adopts text prompts, and the auxiliary information is the second auxiliary information.

[0029] For example, in some specific implementations, the retrieval enhancement subsystem is designed to improve the accuracy and timeliness of information during the large language model generation process, and consists of a local knowledge base retrieval unit and an internet retrieval unit. The local knowledge base retrieval unit retrieves structured information from a pre-built and optimized internal database (local knowledge base), ensuring that the large language model has access to detailed and verified professional knowledge when generating reports. The internet retrieval unit utilizes advanced web crawling and data processing technologies to acquire external information resources in real time, providing the latest industry trends, research progress, and public data to ensure the timeliness and relevance of the generated content; specifically, the internet retrieval unit can retrieve relevant web page links, and then, combined with crawling technology, crawl the full-text content of the corresponding websites as external information resources.

[0030] In some embodiments, the local knowledge base includes a data source system, a retrieval system, and intelligent summarization; the retrieval enhancement subsystem also includes a local knowledge base construction unit, which is specifically used to perform the following operations: The pre-collected data sources are standardized to obtain a data source system; the standardization process includes format unification conversion, text structure parsing, and key information point extraction; the data sources are dynamically updated periodically based on a preset cycle; Based on the processing results of key information point extraction, a retrieval system is constructed using a multi-level knowledge tree structure; the multi-level knowledge tree structure includes subject areas, sub-topics, and specific cases / policies; Based on the retrieval system, a large language model is invoked to generate intelligent summaries step by step from specific cases / policies.

[0031] Exemplary examples, such as in some specific implementations, Figure 2 As shown, the construction process of a local knowledge base can be implemented as follows: 1) Data Source Collection and Organization: Data is extensively collected from various sources, including government agencies, industry research organizations, academic publications, and internal company documents, ensuring comprehensive and authoritative information coverage. To handle information of varying formats and qualities, a unified format conversion is performed, and redundant data is removed to improve information quality. Regular expression rules are used to parse the text structure to extract key information points and establish a logically clear data source system. This step not only lays a solid foundation for subsequent steps but also ensures that all information entering the knowledge base undergoes rigorous screening and standardization.

[0032] 2) Constructing an efficient retrieval system: A multi-level knowledge tree structure is created to organize information. This structure includes, but is not limited to, nodes for subject areas, subtopics, specific cases, or policies, aiming to reflect the inherent hierarchical relationships of information. To achieve efficient storage and retrieval, a tree-structured database system is adopted. Non-leaf nodes are used as indexes and do not directly store specific content; instead, they aggregate information from lower-level nodes to enhance query speed. Leaf nodes are responsible for storing the actual consultation content. This architecture design allows users to quickly locate the information they need while ensuring the system's scalability and flexibility.

[0033] 3) Intelligent Summary Generation: The third step involves applying an advanced large language model to generate intelligent summaries from the most basic specific cases or policies, progressively upwards. This considers the interrelationships between different fields, such as how changes in economic policies affect the operational strategies of specific industries, thus providing richer background information. For each node, whether it's a bottom-level leaf node or a higher-level non-leaf node, a high-quality summary is generated based on the content of its subordinate nodes. For leaf nodes, the summary is generated not only based on the content of the article itself but also includes other content extracted from it through cross-domain references, resulting in a comprehensive summary. For non-leaf nodes, the summary information of all subordinate leaf nodes is aggregated, and a high-level summary is generated using the large language model. The content and summary information of each node are vectorized and stored based on an embedding model to ensure retrieval performance.

[0034] 4) Continuous Updates and Maintenance: Recognizing the importance of keeping the knowledge base up-to-date, a dynamic update mechanism has been established to ensure it reflects the latest market dynamics, regulatory changes, and research findings. Therefore, it is necessary to regularly review and update the information added to the database to ensure its continued effectiveness.

[0035] An automated information update process is achieved by constructing an intelligent data acquisition system. This system pre-sets a series of authoritative information sources, scans each site according to a predetermined cycle (e.g., every three days), and filters and integrates newly published information into the knowledge base based on the relevance of the publication time and title. This process ensures that the knowledge base is always kept in an optimal state, providing the most valuable auxiliary information for subsequent report generation.

[0036] The report outline generation subsystem is used to generate a report outline based on guiding titles and knowledge data, by calling a large language model; the knowledge data includes data retrieved from the internal knowledge base; when auxiliary information exists, the knowledge data also includes auxiliary information. It should be noted that, in some embodiments, the report outline generation subsystem includes a current situation summary generation unit, a problem and challenge analysis unit, and a countermeasures and suggestions generation unit; the report outline includes a current situation summary, problem and challenge analysis data, and countermeasures and suggestions; when the report outline generation subsystem generates a report outline based on guiding titles and knowledge data, and calls a large language model, it is specifically used to perform the following operations: The current situation summary generation unit uses the guiding title and knowledge data to call the large language model to accurately extract the current situation summary. Based on the current situation, the problem and challenge analysis unit summarizes the current situation and uses a large language model for analysis and reasoning to obtain problem and challenge analysis data. The countermeasures and suggestions generation unit analyzes the current situation and challenges based on the data, and then uses a large language model to perform suggestion analysis to obtain countermeasures and suggestions.

[0037] For example, in some specific implementations, the report outline generation subsystem is dedicated to automatically generating a well-structured and detailed article outline based on the article title provided by the user. This subsystem comprises three core components: a current situation summary generation unit, a problem and challenge analysis unit, and a countermeasures and suggestions generation unit.

[0038] The current status summary generation unit accurately extracts and summarizes the current status description of the domain by integrating user-input title information and auxiliary data (if available) provided by the retrieval enhancement subsystem. This process relies not only on static information stored in the local knowledge base but also on the latest developments obtained by the internet retrieval unit, ensuring that the current status summary is both comprehensive and timely.

[0039] The Problems and Challenges Analysis unit, based on the current situation summary, starts with the summarized description of the current situation and utilizes the reasoning capabilities of a large model to deeply analyze existing problems and challenges based on the generated current situation summary. This unit can identify key obstacles affecting the development of the field and provide a scientific basis for subsequent countermeasures and recommendations.

[0040] The policy suggestion generation unit proposes specific and feasible solutions based on a comprehensive summary of the current situation and analysis of the challenges. Through intelligent matching and optimization, it generates corresponding policy suggestions for each identified problem, thus constructing a logically rigorous and well-structured outline. The entire process ensures coherence and consistency from the description of the current situation to the analysis of the problems and the proposal of solutions, ultimately providing users with a clear, informative, and high-quality outline.

[0041] The report text generation subsystem is used to generate the report text based on the report outline and knowledge data by calling a large language model.

[0042] It should be noted that in some embodiments, the report text generation subsystem includes a report text generation unit; when generating the report text based on the report outline and knowledge data by calling the large language model, the report text generation subsystem specifically performs the following operations: The report outline is broken down and transformed to obtain the guiding / topic sentence corresponding to each outline item; Text prompts are generated based on the guiding / topic sentences; Based on text prompts, the system retrieves supporting data from knowledge data, and then calls a large language model to integrate the content and generate the report text corresponding to each outline item.

[0043] For example, in some specific implementations, the report body generation subsystem performs refined content filling based on the structured article outline provided by the report outline generation subsystem. This subsystem first finely decomposes the generated outline, leveraging the superior text analysis capabilities of the large language model to refine each outline item into a paragraph's introductory sentence or topic sentence. Subsequently, new text prompts are constructed based on the transformed sentences, continuously improving sentence quality and expressive effect through an iterative optimization process. This ensures that the beginning of each paragraph accurately conveys the core idea while naturally leading into subsequent content, thus achieving an efficient conversion from outline to fluent text. Following this, a secondary determination of information needs is made. In some optional implementations, to ensure the comprehensiveness of the body text, the efficient information collection capabilities of the retrieval enhancement system can be directly utilized to accurately search and integrate auxiliary information for the first sentence of each paragraph, providing detailed contextual support for the large language model. This ensures that each generated paragraph is both in-depth and coherent. This process not only improves the professionalism and accuracy of the text but also guarantees the logical rigor and clarity of the entire report content.

[0044] In some embodiments, the report body generation subsystem further includes a report summary generation unit, which is specifically used to perform the following operations: Based on all the report texts, the core viewpoints and key conclusions are extracted by using a large language model, and then combined with the report texts to construct prompt texts. Based on the prompt text, the large language model is invoked to generate a report summary.

[0045] For example, in some specific implementations, after the main text is generated, the report summary generation unit immediately starts, automatically generating a concise and comprehensive summary based on the complete main text content. This unit leverages the powerful text understanding capabilities of the large-scale model to accurately extract core viewpoints and key conclusions from the entire report. Subsequently, these extracted key points are combined with the original text to construct a concise prompt text, guiding the large-scale model to generate the report's main text summary. This process ensures that the summary not only accurately captures the essence of the report but also maintains consistency and coherence with the original text. Ultimately, through the organic combination of the main text and the summary, a structurally complete, content-rich, and professional report is formed, meeting the needs of users in different scenarios. This systematic generation process from outline to main text to summary demonstrates the advanced nature and practicality of this invention in the field of automated professional document creation, significantly improving the quality and efficiency of report generation.

[0046] To explain in detail the principle of the technical solution of the present invention, the overall process of the present invention will be described below with reference to some specific embodiments. It is easy to understand that the following is an explanation of the technical principle of the present invention and should not be regarded as a limitation of the present invention.

[0047] First, it should be noted that this invention aims to leverage the highly flexible understanding and generation capabilities of advanced large language models to not only accurately describe the current situation but also identify and analyze existing problems in detail, propose targeted solutions, and support rapid customization to meet the needs of specific industries. This provides a more intelligent, efficient, and customized professional report generation solution, significantly improving the efficiency and quality of users writing complex and important documents.

[0048] In view of this, such as Figure 3 As shown, the present invention provides a professional report generation system based on a large language model, which mainly consists of two parts: a business logic layer and a large language model layer.

[0049] The large language model layer forms the foundation of the entire system. This layer fully leverages the superior text generation and understanding capabilities of the large language model, providing solid technical support for the "Report Outline Generation Subsystem" and "Report Body Generation Subsystem" in the business logic layer. By deeply integrating the powerful functions of the large language model, not only is the accuracy and rationality of these two subsystems in constructing the report structure ensured, but the richness and professionalism of their content creation are also significantly enhanced. As the intelligent core, the large language model enables the system to demonstrate outstanding capabilities in parsing complex information, capturing subtle semantics, and generating coherent and insightful text, thus laying the foundation for ultimately generating high-quality professional reports.

[0050] The business logic layer mainly consists of three parts: Search Enhancement Subsystem: The retrieval enhancement subsystem is designed to improve the accuracy and timeliness of information during the large language model generation process. It consists of a local knowledge base retrieval unit and an internet retrieval unit. The local knowledge base retrieval unit retrieves structured information from a pre-built and optimized internal database (local knowledge base), ensuring that the large language model has access to detailed and verified professional knowledge when generating reports. The internet retrieval unit utilizes advanced web crawling and data processing technologies to acquire external information resources in real time, providing the latest industry trends, research progress, and public data to ensure the timeliness and relevance of the generated content. The construction process of the local knowledge base is as follows: Figure 2 As shown, the specific process details are as described above in the specific implementation method, and will not be repeated here.

[0051] Report outline generation subsystem: The report outline generation subsystem is dedicated to automatically generating well-structured and detailed article outlines based on user-provided article titles. This subsystem comprises three core components: a current situation summary generation unit, a problem and challenge analysis unit, and a countermeasures and suggestions generation unit.

[0052] The current status summary generation unit accurately extracts and summarizes the current status description of the domain by integrating user-input title information and auxiliary data (if available) provided by the retrieval enhancement subsystem. This process relies not only on static information stored in the local knowledge base but also on the latest developments obtained by the internet retrieval unit, ensuring that the current status summary is both comprehensive and timely.

[0053] The Problems and Challenges Analysis unit, based on the current situation summary, starts with the summarized description of the current situation and utilizes the reasoning capabilities of a large model to deeply analyze existing problems and challenges based on the generated current situation summary. This unit can identify key obstacles affecting the development of the field and provide a scientific basis for subsequent countermeasures and recommendations.

[0054] The policy suggestion generation unit proposes specific and feasible solutions based on a comprehensive summary of the current situation and analysis of the challenges. Through intelligent matching and optimization, it generates corresponding policy suggestions for each identified problem, thus constructing a logically rigorous and well-structured outline. The entire process ensures coherence and consistency from the description of the current situation to the analysis of the problems and the proposal of solutions, ultimately providing users with a clear, informative, and high-quality outline.

[0055] Report text generation subsystem: The report body generation subsystem performs refined content filling based on the structured article outline provided by the report outline generation subsystem. This subsystem first meticulously decomposes the generated outline, leveraging the superior text analysis capabilities of the large-scale model to refine each outline item into a paragraph's introductory sentence or topic sentence. Subsequently, new text prompts are constructed based on the transformed sentences, continuously improving sentence quality and expressiveness through iterative optimization. This ensures that the beginning of each paragraph accurately conveys the core idea while naturally leading into subsequent content, achieving an efficient conversion from outline to fluent text. Following this, a secondary determination of information requirements is performed. In some optional implementations, to ensure the comprehensiveness of the body text, the efficient information collection capabilities of the retrieval enhancement system can be directly utilized. For the first sentence of each paragraph, auxiliary information is precisely searched and integrated, providing detailed contextual support to the large-scale language model, thus ensuring that each generated paragraph is both in-depth and coherent. This process not only enhances the professionalism and accuracy of the text but also guarantees the logical rigor and clear structure of the entire report.

[0056] After the main text is generated, the report summary generation unit immediately starts, automatically generating a concise and comprehensive summary based on the complete main text content. Leveraging the powerful text understanding capabilities of the large-scale model, this unit accurately extracts the core viewpoints and key conclusions from the entire report. Subsequently, these extracted key points are combined with the original text to construct a concise prompt text, guiding the large-scale model to generate the report's main text summary. This process ensures that the summary not only accurately captures the essence of the report but also maintains consistency and coherence with the original text. Ultimately, through the organic combination of the main text and the summary, a structurally complete, content-rich, and professional report is formed, meeting the needs of users in different scenarios. This systematic generation process from outline to main text to summary demonstrates the advanced nature and practicality of this invention in the field of automated professional document creation, significantly improving the quality and efficiency of report generation.

[0057] On the other hand, embodiments of the present invention also provide a method for generating professional reports. It is understood that the method for generating professional reports provided by these embodiments can be applied to any computer device with data processing and computing capabilities, and this computer device can be various types of terminals or servers. When the computer device in the embodiments is a server, the server is an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms. Optionally, the terminal can be a smartphone, tablet computer, laptop computer, or desktop computer, but is not limited to these.

[0058] like Figure 4 The diagram shown is a schematic representation of an implementation environment provided by an embodiment of the present invention. (Refer to...) Figure 4 The implementation environment includes at least one terminal 102 and a server 101. The terminal 102 and the server 101 can be connected via a network, either wirelessly or via a wired connection, to complete data transmission and exchange.

[0059] Server 101 can be a standalone physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms.

[0060] Additionally, server 101 can also be a node server in a blockchain network. Blockchain is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms.

[0061] Terminal 102 can be a smartphone, tablet computer, laptop computer, desktop computer, smart speaker, smartwatch, etc., but is not limited to these. Terminal 102 and server 101 can be directly or indirectly connected via wired or wireless communication, and this embodiment of the invention does not impose any limitations.

[0062] Exemplary based on Figure 4 The implementation environment shown in this embodiment of the invention provides a professional report generation method. The following description uses the application of this professional report generation method in terminal 102 as an example. It can be understood that this professional report generation method can also be applied in server 101.

[0063] Reference Figure 5 , Figure 5 The flowchart illustrates a professional report generation method applied to a terminal, as provided in an embodiment of the present invention. The executing entity of this professional report generation method can be any of the aforementioned computer devices (including servers or terminals). (Refer to...) Figure 5 The method may include the following steps: S100: Based on the guiding title and report outline generated by the target object's input and the report outline generated by the subsystem, the large language model is used to analyze and determine whether auxiliary information is needed; if auxiliary information is needed, a calling instruction is generated. S200, The retrieval enhancement subsystem retrieves auxiliary information in response to the call command; S300: The report outline generation subsystem generates a report outline based on guiding titles and knowledge data by calling a large language model; the knowledge data includes data retrieved from the internal knowledge base; when auxiliary information exists, the knowledge data also includes auxiliary information. S400: The report text generation subsystem generates the report text based on the report outline and knowledge data by calling the large language model.

[0064] In some embodiments, the local knowledge base includes a data source system, a retrieval system, and intelligent summarization; the method may also include the following steps: The pre-collected data sources are standardized to obtain a data source system; the standardization process includes format unification conversion, text structure parsing, and key information point extraction; the data sources are dynamically updated periodically based on a preset cycle; Based on the processing results of key information point extraction, a retrieval system is constructed using a multi-level knowledge tree structure; the multi-level knowledge tree structure includes subject areas, sub-topics, and specific cases / policies; Based on the retrieval system, a large language model is invoked to generate intelligent summaries step by step from specific cases / policies.

[0065] In some embodiments, the method may further include the following steps: Based on all the report texts, the core viewpoints and key conclusions are extracted by using a large language model, and then combined with the report texts to construct prompt texts. Based on the prompt text, the large language model is invoked to generate a report summary.

[0066] Exemplary examples, such as in some specific implementations, Figure 6 The diagram shown is the overall flowchart of the report generation system. The specific process can be implemented as follows: 1) The user inputs a guiding title into the system. This title should summarize the main theme and purpose of the report. The user may need to consider the target audience, the scope of the report, and the key messages they want to convey, as the basis for the entire report generation process.

[0067] 2) After receiving the title, the large language model performs semantic analysis to determine the subject area of ​​the report and assess whether additional information is needed to ensure the accuracy and depth of the content. This includes checking whether there is sufficient relevant data in the internal knowledge base or whether there is information that needs to be updated. If it is determined that auxiliary information is needed, the model will activate the retrieval enhancement module. This system uses a text similarity calculation algorithm to calculate the similarity between the article title and text fragments in a pre-built local knowledge base, and obtains text materials related to the title based on the similarity score. To further ensure the timeliness and comprehensiveness of the retrieved information, it also uses internet search engine APIs to retrieve web pages related to the title. Finally, the retrieval results are organized and passed to the large language model for subsequent steps.

[0068] It should be noted that, when the retrieval enhancement subsystem retrieves the auxiliary information in response to a call command, it can also be used to perform the following operations: If the breadth of information coverage of the auxiliary information (including the first auxiliary information and the second auxiliary information) does not meet the preset judgment conditions, it is determined that the auxiliary information does not meet the requirements.

[0069] Specifically, the breadth of information coverage includes data sources, types, and quantities. The breadth can be judged based on pre-set judgment thresholds for each dimension of breadth.

[0070] Then, the large language model is invoked to update the retrieval information (including subject areas and / or text prompts), and the retrieval enhancement subsystem responds again to the invocation command to retrieve auxiliary information.

[0071] For example, given that the auxiliary information obtained through simple command retrieval may be inaccurate or not fully meet the downstream outline generation requirements, this embodiment of the invention introduces a large-scale model at this stage to evaluate the retrieved content. If the information retrieved initially is not comprehensive enough, the system will automatically generate new search commands (including search keywords and phrases) to continue searching iteratively until the large-scale model confirms that the obtained auxiliary information is sufficiently detailed to meet the requirements. This ensures that the final information obtained is both accurate and comprehensive, effectively supporting subsequent task execution.

[0072] 3) Generating the Current Situation Summary: Based on the user-provided title, the large language model begins constructing the first part of the report outline—the current situation summary. This part aims to explain the current context and provide necessary background information. To generate an accurate and comprehensive summary, the model first parses the keywords and thematic terms in the title and, combined with previously retrieved auxiliary information (if any) and its own knowledge base, organizes one or more paragraphs summarizing the current state of the field. In this process, the model considers multiple factors, including but not limited to major achievements, market trends, policy environment, and technological developments, to ensure readers have a comprehensive understanding of the topic under discussion. To ensure the relevance and authority of the content, the model selects the most relevant and up-to-date information from multiple sources, evaluates it using text similarity calculation algorithms and semantic analysis techniques, and integrates information from the local knowledge base, academic literature, industry reports, and news reports to form a coherent whole. Finally, to verify the accuracy of the summary, the model's reasoning capabilities are used to check the accuracy of each sentence in the paragraphs and provide a verification conclusion, thereby ensuring the accuracy of the generated result.

[0073] 4) Identifying Problems and Challenges: Building upon the current situation summary, the large language model conducts further in-depth analysis to identify the key problems and challenges currently facing the system. The task at this stage is to pinpoint obstacles hindering the continued development of the current situation or potential future difficulties. The model utilizes logical reasoning and critical thinking skills, combined with information from the current situation summary, to propose a series of specific issues. These problems and challenges will become the core topics of the subsequent policy recommendations section and the focus of the entire report.

[0074] It should be noted that after the problem and challenge analysis unit summarizes the current situation and uses a large language model for analysis and reasoning to obtain problem and challenge analysis data, the report outline generation subsystem can also be used to perform the following operations: Based on the subject domain, a large language model is invoked to perform similarity matching on the Internet or a pre-built local knowledge base to obtain similarity report data; New identification data for problem challenges were extracted from similar report data; Newly identified problem challenge data is integrated into the problem challenge analysis data to update the problem challenge analysis data.

[0075] For example, after generating a problem and challenge outline, embodiments of the present invention can also retrieve report data similar to the article's topic from the internet or a local knowledge base, and extract questions and challenges raised by human experts for such situations. The system then integrates these existing questions with the newly identified questions and challenges, further generating more in-depth and comprehensive questions and challenges based on the insights of human experts. This method not only ensures a deep understanding of the current issue but also promotes the development of innovative thinking, providing new perspectives for solving complex problems.

[0076] 5) Developing Recommendations: With the problems and challenges clearly defined, the large language model now focuses on proposing feasible solutions. Recommendations should be specific, actionable, and take into account the complexity of the actual situation. The model will propose different levels of responses based on the nature of the problem, ranging from short-term actions to long-term strategic planning. In this way, the model ensures that each proposed solution effectively addresses the current challenges and lays the foundation for continuous improvement in the future.

[0077] It should be noted that after the countermeasures and suggestions generation unit obtains countermeasures and suggestions by summarizing the current situation and analyzing the problem and challenges through data, and by calling the large language model for suggestion analysis, the report outline generation subsystem can also be used to perform the following operations: The large language model is invoked to perform feasibility analysis on the proposed countermeasures. If any of the countermeasures are deemed unfeasible, the problem challenge analysis data corresponding to the unfeasible countermeasures is retrieved by the enhancement subsystem to retrieve solution strategies. The countermeasures are then updated based on the retrieval results.

[0078] For example, after the model generates policy recommendations, to ensure these recommendations are highly feasible, the system will again invoke the model to conduct in-depth analysis of the generated results to verify their feasibility. If some recommendations are found to be infeasible, the system will automatically activate the search module to find solutions proposed by human experts for such problems. Then, based on these expert strategies and the actual situation, the model will propose new, more feasible solutions.

[0079] 6) Second Assessment of Supporting Information Needs: Before writing the main text, to ensure the large model can fully utilize more comprehensive and accurate supporting information during the generation process, the model will conduct a second assessment based on the generated report outline to determine whether additional supporting information is needed to support detailed arguments. This process aims to enhance the depth and accuracy of the content, ensuring that each paragraph is adequately supported and argued. This assessment not only focuses on the overall accuracy of the report but also on the support provided for specific chapters. For example, some policy recommendations may require citing the latest research findings or statistical data. If necessary, the retrieval enhancement module will be activated again to collect the required materials. Subsequently, this information will be integrated into the corresponding main text sections to ensure the arguments are sufficiently strong.

[0080] 7) Writing the Report Body: Following the completed outline, the large language model begins writing the complete report body. During the writing process, the model emphasizes maintaining logical coherence and narrative fluency, while ensuring that each section flows naturally. To enhance the report's quality, the model also employs appropriate rhetorical devices and expressions to make the article more engaging. Furthermore, the model carefully controls the length, avoiding redundancy or digressions.

[0081] 8) Writing the Report Summary: After the main body of the report is completed, the large language model will summarize the entire report, extract the most important viewpoints and conclusions, and write a concise summary. This summary is a high-level overview of the entire report, aiming to allow readers to understand the core content and conclusions of the report in a short time. The summary should include the purpose, methods, main findings, and final recommendations of the report, but should not go into too much detail so that readers can quickly grasp the key points.

[0082] 9) Output the complete report: The final step is to combine all components—title, outline, body, and abstract—into a single, complete document. The model ensures consistent formatting, aesthetically pleasing layout, and compliance with user requirements (such as font size and margins). Once everything is ready, the report will be output as an electronic file for users to download, print, or view online. The system also offers options for users to choose different output formats (such as PDF and Word) to meet diverse usage needs.

[0083] It should be noted that the content of the system embodiments of the present invention is applicable to the method embodiments of the present invention. The specific process steps implemented in the method embodiments of the present invention are the same as those in the above system embodiments, and the beneficial effects achieved are also the same as those achieved by the above systems. It should be understood that the functional logic implemented by the specific modules of each part of the above system embodiments all fall within the protection scope of the method embodiments of the present invention, and will not be described in detail here.

[0084] On the other hand, embodiments of the present invention also provide an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the aforementioned professional report generation method. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0085] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0086] like Figure 7 As shown, Figure 7 This illustration shows a specific example of the hardware structure of an electronic device 1000 according to one embodiment. The electronic device 1000 includes: The processor 1001 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present invention. The memory 1002 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 1002 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1002 and is called and executed by the processor 1001 to execute the network node population optimization method of the embodiments of this invention. Input / output interface 1003 is used to implement information input and output; The communication interface 1004 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 1005 transmits information between various components of the device (e.g., processor 1001, memory 1002, input / output interface 1003, and communication interface 1004); The processor 1001, memory 1002, input / output interface 1003 and communication interface 1004 are connected to each other within the device via bus 1005.

[0087] The electronic device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0088] The content of the method embodiments of the present invention is applicable to the embodiments of the present electronic device. The specific functions implemented by the embodiments of the present electronic device are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above methods.

[0089] Another aspect of this invention provides a computer-readable storage medium storing a program that is executed by a processor to implement the aforementioned method.

[0090] It should be noted that the computer-readable medium shown in the embodiments of the present invention can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disc read-only memory (CD to ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, wherein computer-readable program code is carried. Such transmitted data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wired, etc., or any suitable combination thereof.

[0091] The content of the method embodiments of the present invention is applicable to the computer-readable storage medium embodiments. The specific functions implemented by the computer-readable storage medium embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above methods.

[0092] This invention also discloses a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium and execute the computer instructions, causing the computer device to perform the aforementioned method.

[0093] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0094] It should be noted that although several modules for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to embodiments of the present invention, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0095] Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, portable hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, touch terminal, or network device, etc.) to execute the method according to the embodiments of the present invention.

[0096] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order shown in the operation diagrams. For example, depending on the functions / operations involved, two consecutively shown blocks may actually be executed substantially simultaneously, or the blocks may sometimes be executed in reverse order. Furthermore, the embodiments presented and described in the flowcharts of this invention are provided by way of example to provide a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented herein. Alternative embodiments are contemplated in which the order of various operations is changed and sub-operations described as part of a larger operation are executed independently.

[0097] Furthermore, although the invention has been described in the context of functional modules, it should be understood that, unless otherwise stated, one or more of the functions and / or features may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding the invention. Rather, given the properties, functions, and internal relationships of the various functional modules in the apparatus disclosed herein, the actual implementation of the module will be understood within the scope of conventional skill of an engineer. Therefore, those skilled in the art can implement the invention as set forth in the claims using ordinary techniques without excessive experimentation. It is also understood that the specific concepts disclosed are merely illustrative and not intended to limit the scope of the invention, which is determined by the full scope of the appended claims and their equivalents.

[0098] If a function is implemented as 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 this invention, or the part that contributes to the prior art or the part of this invention, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0099] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution means, apparatus, or device (such as a computer-based device, a processor-including device, or other means that can fetch and execute instructions from, or in conjunction with, an instruction execution means, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution means, apparatus, or device.

[0100] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0101] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution device. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0102] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0103] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

[0104] The above is a detailed description of the preferred embodiments of the present invention. However, the present invention is not limited to the embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications or substitutions are included within the scope defined by the claims of the present invention.

Claims

1. A professional report generation system, characterized in that, It includes a large language model layer and a business logic layer, wherein the business logic layer includes a retrieval enhancement subsystem, a report outline generation subsystem, and a report body generation subsystem; The large language model layer is pre-set with a large language model; The report outline generation subsystem is used to generate a report outline based on the guiding title input by the target object and combined with knowledge data, by calling the large language model; wherein, the knowledge data includes retrieval data from the internal knowledge base; The large language model is used to retrieve first retrieval data from the internal knowledge base based on the guiding title before the report outline generation subsystem generates the report outline. When the amount of the first retrieval data meets the preset conditions, it is determined that the first auxiliary information is not needed; otherwise, it is determined that the first auxiliary information is needed and a first calling instruction is generated. The retrieval enhancement subsystem is used to retrieve the first auxiliary information in response to the first invocation command, and to incorporate the first auxiliary information into the knowledge data; The large language model is also used to construct text prompts based on the report outline before the report text generation subsystem generates the report text, retrieve second retrieval data from the internal knowledge base through the text prompts, and determine that second auxiliary information is not needed when the amount of second retrieval data meets the preset conditions; otherwise, it determines that second auxiliary information is needed and generates a second calling instruction. The retrieval enhancement subsystem is also configured to retrieve the second auxiliary information in response to the second invocation instruction, and to incorporate the second auxiliary information into the knowledge data; The report text generation subsystem is used to generate the report text based on the report outline and the knowledge data by calling the large language model.

2. The professional report generation system according to claim 1, characterized in that, When the large language model retrieves the first retrieval data from the internal knowledge base based on the guiding title, it is specifically used to perform the following operations: Semantic analysis of the guiding titles yields the subject areas; The first retrieval data is obtained from the internal knowledge base based on the stated subject area.

3. The professional report generation system according to claim 2, characterized in that, The retrieval enhancement subsystem includes a local knowledge base retrieval unit and an internet retrieval unit; when the retrieval enhancement subsystem retrieves the first auxiliary information in response to the first invocation instruction, it is specifically used to perform the following operations: In response to the first invocation instruction, according to the topic domain, structured information is retrieved from a pre-built local knowledge base through the local knowledge base retrieval unit, and external information resources are obtained through web crawling through the Internet retrieval unit; The first auxiliary information is obtained by integrating the structured information and the external information resources.

4. The professional report generation system according to claim 3, characterized in that, The local knowledge base includes a data source system, a retrieval system, and intelligent summarization; the retrieval enhancement subsystem further includes a local knowledge base construction unit, which is specifically used to perform the following operations: The pre-collected data sources are standardized to obtain a data source system; the standardization process includes format unification conversion, text structure parsing, and key information point extraction; the data sources are dynamically updated periodically based on a preset cycle; Based on the processing results of the key information points extracted, a retrieval system is constructed using a multi-level knowledge tree structure; the multi-level knowledge tree structure includes subject areas, sub-topics, and specific cases / policies; Based on the retrieval system, the large language model is invoked to generate intelligent summaries step by step from the specific case / policy upwards.

5. The professional report generation system according to claim 1, characterized in that, The report outline generation subsystem includes a current situation summary generation unit, a problem and challenge analysis unit, and a countermeasures and suggestions generation unit; the report outline includes a current situation summary, problem and challenge analysis data, and countermeasures and suggestions; when the report outline generation subsystem generates a report outline based on the guiding title input by the target object and combined with knowledge data, it is specifically used to perform the following operations: The current situation summary generation unit uses the guiding title and the knowledge data to call the large language model to accurately extract the current situation summary. Based on the current situation, the problem and challenge analysis unit summarizes the current situation and uses the large language model to perform analysis and reasoning to obtain problem and challenge analysis data. Based on the current situation summary and the problem challenge analysis data, the countermeasure suggestion generation unit calls the large language model to perform suggestion analysis and obtain countermeasure suggestions. Specifically, after obtaining the problem challenge analysis data through analysis and reasoning, the report outline generation subsystem is further used to perform the following operations: Based on the topic domain corresponding to the guiding title, a large language model is invoked to perform similarity matching on the Internet or a pre-built local knowledge base to obtain similarity report data; New problem and challenge identification data were extracted from similar report data; Integrate newly identified problem challenge data into the problem challenge analysis data to update the problem challenge analysis data; After obtaining the proposed countermeasures through analysis, the report outline generation subsystem is further configured to perform the following operations: The large language model is invoked to perform feasibility analysis on the proposed countermeasures. If any of the countermeasures are deemed unfeasible, the problem challenge analysis data corresponding to the unfeasible countermeasures is retrieved by the enhancement subsystem to retrieve solution strategies. The countermeasures are then updated based on the retrieval results.

6. The professional report generation system according to claim 1, characterized in that, The report text generation subsystem includes a report text generation unit; when generating the report text based on the report outline and the knowledge data by calling the large language model, the report text generation subsystem specifically performs the following operations: The report outline is decomposed and transformed to obtain the guiding / topic sentence corresponding to each outline item; Text prompts are constructed based on the aforementioned guiding / topic sentences; Based on the text prompts, the supporting data for the main text is retrieved from the knowledge data, and then the large language model is invoked to integrate the content and generate the report text corresponding to each outline item.

7. The professional report generation system according to claim 6, characterized in that, The report body generation subsystem also includes a report summary generation unit, which is specifically used to perform the following operations: Based on all the report texts, the core viewpoints and key conclusions are extracted by calling the large language model, and then combined with the report texts to construct prompt texts. Based on the prompt text, the large language model is invoked to generate a report summary.

8. A method for generating professional reports, characterized in that, Applied to the professional report generation system as described in any one of claims 1 to 7, the method includes the following steps: The report outline generation subsystem generates a report outline by combining the guiding title input by the target object with knowledge data and calling a large language model; wherein, the knowledge data includes retrieval data from an internal knowledge base; Before the report outline generation subsystem generates the report outline, the large language model retrieves the first retrieval data from the internal knowledge base based on the guiding title. When the amount of the first retrieval data meets the preset conditions, it is determined that the first auxiliary information is not needed; otherwise, it is determined that the first auxiliary information is needed and a first calling instruction is generated. The retrieval enhancement subsystem retrieves the first auxiliary information in response to the first invocation command and incorporates the first auxiliary information into the knowledge data. Before the report body generation subsystem generates the report body using the large language model, a text prompt is constructed based on the report outline. Second retrieval data is retrieved from the internal knowledge base using the text prompt. If the amount of the second retrieval data meets a preset condition, it is determined that the second auxiliary information is not needed; otherwise, it is determined that the second auxiliary information is needed and a second calling instruction is generated. The retrieval enhancement subsystem retrieves the second auxiliary information in response to the second invocation command and incorporates the second auxiliary information into the knowledge data. The report text generation subsystem is used to generate the report text based on the report outline and the knowledge data by calling the large language model.

9. An electronic device, characterized in that, Including the processor and memory; The memory is used to store programs; The processor executes the program to implement the method as described in claim 8.

10. A computer storage medium storing a processor-executable program, characterized in that, The processor-executable program, when executed by the processor, is used to implement the method as described in claim 8.

Images