9 results about "Report writing" patented technology

The invention belongs to the technical field of shipping big data application, and particularly relates to a shipping research report generation method based on multi-agent cooperation, and the method comprises the steps: enabling a multi-agent hierarchical cooperation architecture, dividing a whole agent into a plurality of sub-agents with independent functions, and dividing the sub-agents into a management layer and an execution layer; the management layer comprises a project manager agent, an architect agent and a demand side agent; the execution layer comprises a shipping intention recognition agent, a background investigation agent, a content team agent, a review verification agent and a report writing agent; the intelligent agents cooperate with one another, and a shipping research report is written in a popular, easy-to-understand, organized and clear language according to the structure and style of the shipping report. According to the method provided by the invention, human information processing, decision making and operation processes are simulated, a modularized and multi-agent cooperative agent scheme is constructed, and the application of AI in the shipping field is pushed to a new height of comprehensive knowledge creation from operation assistance and single analysis.

The invention provides a financial research report generation method based on a large model. Comprising a data collection module, a data analysis module and a report writing module. The data collection module is used for receiving a predefined data collection task and an autonomously generated data collection task, then inputting the tasks into a deep search agent, and constructing a structured multi-modal memory for a search result for storage; the data analysis module constructs a data analysis agent, and realizes the generation of a structured multi-modal small report through cooperation of a financial calculation tool, a chart generation tool and a reference addition tool in an actor-reviewer iteration mode in which a large language model is matched with a visual language model; and the report writing module constructs a report writing agent to obtain a chapter first draft and then inputs the chapter first draft into the post-processing agent to obtain a final report. The problems that an existing deep research system and a large language model agent in the financial field are lack of financial field knowledge, limited in multi-modal ability, limited in research and report generation depth and comprehensiveness and the like are solved.

This invention discloses an intelligent research report writing method based on a large language model, comprising four steps: PDF parsing, RAG-based retrieval and recall, LLM-based text generation and polishing, and content numerical verification. The input data for the method model is the original periodic announcement documents. PDF parsing ensures that the text, table, and image information of the announcements can be efficiently identified and used. Next, for the massive amount of information, a RAG-based retrieval and recall scheme is adopted. First, scalar filtering is performed, then a vectorized model is used to hierarchically recall chapter and paragraph information. Then, a rerank model is used to further rank the recalled content, placing more relevant segments at the top for the large model to generate content. The polishing model is specifically trained on historical research report corpora to produce more professional and fluent results. Finally, the content numerical verification section effectively solves the potential errors and risks caused by the illusion of a large model.

The invention provides an AIGC-based financing and leasing industry analysis report writing method and system. The method comprises the following steps: collecting and preprocessing multi-source heterogeneous data; a step of constructing and updating a dynamic knowledge graph; a step of AIGC model training and fine tuning; a complex statistical model integration step; a step of generating a multi-mode content; reinforcing learning optimization report logic; a step of real-time data incremental learning; performing domain expert interactive verification; a step of personalized customization of the report; and carrying out automatic distribution and feedback collection. According to the method, report writing efficiency is remarkably improved, and labor cost is reduced; the report depth is enhanced through an advanced analysis model, and accurate decision making is supported; content innovation is ensured by the dynamic knowledge graph and reinforcement learning, and homogenization is avoided; the timeliness and reliability of the report are improved through real-time data updating and interactive verification; on the whole, the financial lease industry is promoted to be transformed to intelligence and data drive, and the enterprise competitiveness is enhanced.

The application belongs to the technical field of data processing, and specifically discloses an intelligent report generation and deep analysis method for government administration personnel analysis, which comprises the following steps: first, forming information data packets from the relevant data of target personnel; then, inputting the unstructured data in the data packets into a large language model for knowledge extraction and forming a personnel knowledge graph; then, generating corresponding language paragraphs according to the data in the personnel knowledge graph and filling them into a report template; based on the generated report content, question answering, correlation search and thinking chain analysis can also be performed. In view of the problem of low work efficiency of personnel analysis and research for government administration key personnel, the application realizes template-based automatic generation of reports by querying aggregated data and automatically extracting and integrating information by a large model, thereby solving the problem of low work efficiency of government personnel caused by the difficulty in processing unstructured data and the heavy burden of report writing in the personnel analysis and research of government administration key personnel under the prior art.

The application discloses a research report generation method and electronic equipment, relates to the computer technical field, carries out multi-modal content structured extraction to the uploaded document, forms a structured document tree and stores into a vector database; the document key information is retrieved from the database, the document key information is filled into the corresponding task decomposition template, and the research outline is generated after adaptive adjustment; knowledge gap is identified, the optimal execution scheme is determined in combination with the task priority and information demand characteristics; the task is decomposed and iterative retrieval is executed, information is retrieved from the database and external resources to fill the knowledge gap, and the iterative convergence condition is judged through information entropy change and multi-round gain threshold; after judging the iterative convergence, the analysis results generated by the multi-round iteration are integrated and optimized in stages. In this way, the document analysis and report writing efficiency can be improved, the research report with deep thinking is output, the knowledge gap is actively filled, the human limitations are avoided, and the analysis results are comprehensive and accurate.

The invention discloses a periodic research report writing method and device, equipment, a storage medium and a product, and relates to the technical field of data processing.The method comprises the steps that in response to a research report writing request, a historical periodic research report example text is obtained from a database according to parameters in the research report writing request, and the parameters comprise a time range and a research report type; inputting the historical periodic research report example text into the large language model, extracting a research report composition outline, and performing content chapter composition on the research report composition outline to obtain a to-be-evaluated research report; performing reflection modification on the to-be-evaluated research report, and performing research report professional evaluation on the modified to-be-evaluated research report; and if the evaluation result of the research report professional evaluation is that the preset requirement is not met, performing iterative modification and re-evaluation on the to-be-evaluated research report based on the large language model until the evaluation result meets the preset requirement, and outputting the to-be-evaluated research report as a target periodic research report. The problem that high customization and strict specifications are difficult to meet at the same time in the prior art is solved.

The application discloses a property operation report generation method and device based on multiple agents and related media, which comprises the following steps: calling an intention recognition agent to perform intention analysis, and then calling a task planning agent to perform path construction and scheme solving; determining the execution sequence of the task in combination with the task planning data to obtain task processing data; calling a data collection agent to collect data on the task processing data to obtain preliminary processing data; calling a code calculation agent to perform code generation and model operation on the preliminary processing data, then calling an outline making agent to perform outline arrangement and report generation processing, and then calling a report writing agent to complete report packaging, and outputting a property operation report. The application performs outline arrangement and report generation processing on the structured analysis result obtained through calculation, packages and outputs the property operation report, so that the report arrangement is not required to be manually performed, the process is simplified, and the report output efficiency is improved.

The invention relates to an interactive controllable HSE report auxiliary generation method based on a large model, and is suitable for the technical field of artificial intelligence. The method comprises the following steps: obtaining related materials of a report to be generated uploaded by a user, wherein the related materials comprise an HSE notification file, a report template file and a previous high-quality report; calling LLM, extracting multi-level item information from the HSE notification file, extracting a report writing framework from the report template file, and constructing a hierarchical report outline; displaying the report outline to the user, obtaining an editing operation of the user on the report outline, and generating a final outline draft in response to a final draft instruction of the user; splitting a previous high-quality report into chapter blocks according to chapters of the previous high-quality report, and matching each chapter block with each item title in a final outline draft according to semantic similarity to serve as a reference of a writing style during item writing; and calling LLM, generating contents of each chapter of the report based on each item information in the final outline draft and the matched chapter blocks in the previous high-quality report, and synthesizing an HSE report.