Method and system for generating enterprise operation analysis report based on large model

Abstract

The invention discloses a method and system for generating an enterprise operation analysis report based on a large model, and the method comprises the steps: obtaining document data related to the operation of a target unit, and analyzing the document data into Markdown format data suitable for the large model; related prompt pre-stored in a database is obtained; on the basis of the prompt, loading the data in the Markdown format, and assembling the data into a problem proposed for the large model; according to the configured concurrency number, based on the questions proposed by the large model, starting multiple threads to ask questions to the large model; extracting information on the basis of the questions of the large model through the large model, and generating a question and answer result; and filling the question and answer result into an operation analysis report template file of the target unit, and generating an operation analysis report in a preset format.

Description

Technical Field

[0001] The present invention relates to the technical field of information technology applications, and more specifically, to a method and system for generating an enterprise operation analysis report based on a large model. Background Art

[0002] Enterprise operation analysis plays an important role in supervising and guiding enterprise operation decisions. With the complexity of the enterprise operation environment, enterprise operation analysis needs to consider various operation data, including financial data, operation data, credit risk data, tax data, etc. These data reflect the current situation of enterprise operation from multiple aspects. How to accurately evaluate the operation health status of an enterprise from these data has a significant impact on the enterprise value.

[0003] In the field of enterprise operation analysis, currently, data is generally collected manually or based on fixed information processing logic to achieve data statistics, but there are still many limitations and deficiencies. First, the risk of manual data analysis is prone to overlooking and missing potential risks. Especially when dealing with large-scale and complex financial data, key data may be ignored. Second, information processing means can only statistically process data in a fixed number of dimensions, and various dynamic data cannot be processed. It can only be continuously expanded by means of customized development, with low efficiency.

[0004] Prior Art 1 (Application: CN117453919A) provides a method, device, and storage medium for generating a review analysis report based on a large language model. CN117453919A obtains note evaluations through web crawler technology and makes sentiment judgments on the evaluations based on the large language model, which has nothing to do with the enterprise operation analysis report.

[0005] Prior Art 2 (Application: CN118246749A) provides a method and system for financial data risk analysis based on a large model proxy. CN118246749A focuses on risk analysis technology and constructs a specific vector database based on historical data to analyze risks, and cannot generate an analysis report of an enterprise through operation data.

[0006] Therefore, a technology is needed to achieve the generation of an enterprise operation analysis report based on a large model. Summary of the Invention

[0007] The technical solution of the present invention provides a method and system for generating an enterprise operation analysis report based on a large model to solve the problem of how to generate an enterprise operation analysis report based on a large model.

[0008] To solve the above problems, the present invention provides a method for generating an enterprise operation analysis report based on a large model, and the method includes:

[0009] Obtain the document data related to the operation of the target entity, and parse the document data into markdown format data adapted to the large model;

[0010] Obtain the relevant prompts pre-stored in the database;

[0011] Based on the prompt, by loading the markdown format data, assemble it into a question to be asked to the large model;

[0012] According to the configured concurrency number, based on the question asked to the large model, start multi-threaded questioning to the large model;

[0013] Through the large model, extract information based on the question asked to the large model and generate a Q&A result;

[0014] Fill the Q&A result into the operation analysis report template file of the target entity to generate an operation analysis report in a preset format.

[0015] Preferably, it further includes: training the large model for information extraction through a training dataset.

[0016] Preferably, the information extraction includes: named entity recognition, relationship extraction, and event extraction.

[0017] Preferably, the process of extracting information based on the question asked to the large model by the large model and generating a Q&A result includes:

[0018] Formulate a prompt based on the business scenario, and convert the extracted information into Json data for generating charts based on the formulated prompt.

[0019] Preferably, after filling the Q&A result into the operation analysis report template file of the target entity to generate an operation analysis report in a preset format, it includes:

[0020] Split the operation analysis report to obtain the segmented reports after splitting;

[0021] Configure the number of concurrent threads based on the segmented reports after splitting and the computing power information;

[0022] Output the segmented reports based on the requests of multiple configured threads simultaneously.

[0023] Preferably, the process of filling the Q&A result into the operation analysis report template file of the target entity to generate an operation analysis report in a preset format further includes:

[0024] A pre-set Word template, the Word template is pre-arranged with the style and basic content of the template in advance based on the variable placeholder method, and a customized business analysis report is quickly generated through the variable replacement method;

[0025] Convert the business analysis report of the Word template into a PDF version and provide it for the user side to download.

[0026] Based on another aspect of the present invention, the present invention provides a system for generating an enterprise business analysis report based on a large model, and the system includes:

[0027] An initial unit, configured to obtain document data related to the operation of a target unit, parse the document data into markdown format data adapted to the large model; obtain relevant prompts pre-stored in the database;

[0028] An assembly unit, configured to assemble, based on the prompt, by loading the markdown format data, into a question posed to the large model;

[0029] A question-asking unit, configured to, according to the configured concurrency number, based on the question posed to the large model, start multi-threaded question-asking to the large model;

[0030] A generating unit, configured to extract information based on the question posed to the large model by the large model and generate a question-and-answer result;

[0031] A result unit, configured to fill the question-and-answer result into the business analysis report template file of the target unit to generate a business analysis report in a preset format.

[0032] Preferably, the initial unit is further configured to: perform information extraction training on the large model through a training data set.

[0033] Preferably, the information extraction includes: named entity recognition, relationship extraction, and event extraction.

[0034] Preferably, the generating unit, configured to extract information based on the question posed to the large model by the large model and generate a question-and-answer result, is further configured to:

[0035] Formulate a prompt based on the business scenario, and convert the extracted information into Json data for generating charts based on the formulated prompt.

[0036] Preferably, the result unit, configured to fill the question-and-answer result into the business analysis report template file of the target unit to generate a business analysis report in a preset format, is further configured to:

[0037] Split the business analysis report to obtain the segmented reports after splitting;

[0038] Configure the number of concurrent threads based on the segmented report and computing power information after splitting;

[0039] Output the segmented report based on the requests of multiple configured threads simultaneously.

[0040] Preferably, the result unit is used to fill the Q&A result into the business analysis report template file of the target unit to generate a business analysis report in a preset format, and is further used for:

[0041] Pre-set a word template, where the word template arranges the style and basic content of the template in advance based on the variable placeholder method, and quickly generates a customized business analysis report through variable replacement;

[0042] Convert the business analysis report of the word template into a pdf version and provide it for the user side to download.

[0043] The technical solution of the present invention provides a method and system for generating a business analysis report of an enterprise based on a large model. The method includes: obtaining document data related to the operation of the target unit, parsing the document data into markdown format data adapted to the large model; obtaining relevant prompts pre-stored in the database; based on the prompts, assembling the questions to be asked to the large model by loading the markdown format data; according to the configured concurrency number, asking questions to the large model by starting multiple threads based on the questions asked to the large model; extracting information by the large model based on the questions asked to the large model to generate Q&A results; filling the Q&A results into the business analysis report template file of the target unit to generate a business analysis report in a preset format. The technical solution of the present invention is based on the business data uploaded by the enterprise, and realizes multi-dimensional information extraction and summary based on large model fine-tuning and prompts to generate an analysis report. The technical solution of the present invention provides a method and system for generating a business analysis report of an enterprise based on a large model. Through various data provided by the enterprise, it realizes data analysis with the help of the large model, forms a business analysis report with high readability, improves the accuracy and speed of generating the business analysis report, and enhances the enterprise's business prediction ability. Description of the Drawings

[0044] By referring to the following drawings, the exemplary embodiments of the present invention can be more completely understood:

[0045] Figure 1 It is a flowchart of a method for generating a business analysis report of an enterprise based on a large model according to a preferred embodiment of the present invention;

[0046] Figure 2 It is a flowchart of data processing according to a preferred embodiment of the present invention; and

[0047] Figure 3The structural diagram of a system for generating an enterprise operation analysis report based on a large model according to a preferred embodiment of the present invention. Specific embodiments

[0048] Now, exemplary embodiments of the present invention will be described with reference to the accompanying drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. These embodiments are provided to disclose the present invention in detail and completely, and to fully convey the scope of the present invention to those skilled in the art. The terms in the exemplary embodiments shown in the drawings are not intended to limit the present invention. In the drawings, the same unit / element is denoted by the same reference numeral.

[0049] Unless otherwise specified, the terms (including scientific and technical terms) used herein have the ordinary meaning understood by those skilled in the art. Additionally, it can be understood that the terms defined in the commonly used dictionary should be understood to have a meaning consistent with the context of their related fields, and should not be understood as idealized or overly formal meanings.

[0050] Figure 1 The flowchart of a method for generating an enterprise operation analysis report based on a large model according to a preferred embodiment of the present invention.

[0051] The present invention provides a method for information extraction based on a large model. The purpose of information extraction is to extract structured knowledge, such as entities, relationships, and events, from natural language texts. Information extraction generally includes: named entity recognition, relationship extraction, and event extraction. Large models are very capable in aspects such as text understanding. The training of information extraction based on large models is mainly achieved through few-shot learning. Usually, the challenges faced by few-shot learning include overfitting and difficulty in capturing complex relationships. By expanding the parameter scale of LLMs to endow generalization ability, the performance in few-shot scenarios is improved, and specific scenario prompts are formulated to achieve information extraction. For example:

[0052] Table 1 Example of information extraction prompt

[0053]

[0054]

[0055] In the enterprise operation analysis report, a large number of charts are involved, including common bar charts, line charts, and pie charts, and comparative analysis of multi-dimensional data is involved. The present invention provides a chart output method based on a large model. After few-shot learning, Json data required for chart generation is achieved by formulating prompts according to business scenarios. For example:

[0056] Table 2 Example of chart generation prompt

[0057]

[0058]

[0059]

[0060] A large amount of content is involved in the enterprise operation analysis report. Generally, it is divided into multiple chapters or topics. However, the output of existing large models is limited by the number of tokens, generally ranging from 4K to 8K in size, and it is impossible to output a complete operation analysis report at one time, and the time spent on generating the output at one time is too long. The present invention solves the problems of token limitation and long time consumption by splitting the report content, obtaining the report content in segments, and configuring the number of concurrent threads according to the computing power situation, and making multi-threaded requests simultaneously. In addition, the prompts for each topic are configured in the database and can be dynamically maintained, realizing the dynamic requirements of the analysis report.

[0061] Generating a complete enterprise operation analysis report not only involves content, but also layout and styles, and includes mixed content such as tables, charts, and text. The present invention adopts the scheme of presetting a word template. In the template, according to business needs, the styles and basic content of the template are pre-arranged based on the variable placeholder method, and a customized analysis report is quickly generated through the variable replacement method. After the report is generated, the generated word is converted into pdf and provided for the user side to download.

[0062] As Figure 1 shown, the present invention provides a method for generating an enterprise operation analysis report based on a large model. The method includes:

[0063] Step 101: Obtain the document data related to the operation of the target unit, and parse the document data into markdown format data adapted to the large model;

[0064] The user side of the present invention uploads the document data related to the enterprise operation, including but not limited to financial documents (such as balance sheets, income statements, profit statements), tax documents, and risk documents, generally in excel document format;

[0065] The present invention parses the documents uploaded by the user and parses them into markdown format data required by the large model.

[0066] Step 102: Obtain the relevant prompts pre-stored in the database;

[0067] The present invention reads the relevant prompts pre-stored in the database.

[0068] Step 103: Based on the prompts, assemble the questions to be asked to the large model by loading the markdown format data;

[0069] According to the configuration in the prompt read in the above steps, the present invention loads the markdown format data and assembles it into a question posed to the large model.

[0070] Step 104: According to the configured concurrency number, based on the question posed to the large model, start multiple threads to ask questions to the large model;

[0071] Preferably, it further includes: performing information extraction training on the large model through a training data set.

[0072] Step 105: Extract information through the large model based on the question posed to the large model and generate a Q&A result;

[0073] The present invention reads the configured concurrency number, uses the posed question, starts multiple threads to ask questions to the large model, and blocks and receives the streaming return result to obtain the model answer.

[0074] Preferably, the information extraction includes: named entity recognition, relationship extraction, and event extraction.

[0075] Preferably, extracting information through the large model based on the question posed to the large model and generating a Q&A result includes:

[0076] Formulating a prompt based on the business scenario, and converting the extracted information into Json data for generating charts based on the formulated prompt.

[0077] Step 106: Fill the Q&A result into the business analysis report template file of the target unit to generate a business analysis report in a preset format.

[0078] Preferably, after filling the Q&A result into the business analysis report template file of the target unit to generate a business analysis report in a preset format, it includes:

[0079] Splitting the business analysis report to obtain the split sub-reports;

[0080] Configuring the number of concurrent threads based on the split sub-reports and computing power information;

[0081] Outputting the sub-reports based on the configuration of multiple threads requesting simultaneously.

[0082] Preferably, filling the Q&A result into the business analysis report template file of the target unit to generate a business analysis report in a preset format further includes:

[0083] Pre-setting a word template, the word template pre-arranges the style and basic content of the template based on the variable placeholder method, and quickly generates a customized business analysis report through the variable replacement method;

[0084] Convert the business analysis report in the word template into a pdf version and provide it for users to download.

[0085] The present invention reads the business analysis report template file, fills the answers based on the model into the template file, and generates a business analysis report in word format;

[0086] The present invention converts the word format file in the above step into pdf format through a program;

[0087] The present invention returns the result, returns the pdf format document to the front-end request, and completes the whole process. As Figure 2 shown.

[0088] The present invention realizes information extraction and chart generation of the large model through few-shot learning; the present invention realizes the rapid generation and dynamic adjustment of the analysis report through segment generation and prompting engineering; the present invention realizes personalized customization of the analysis report based on the word template.

[0089] The method for generating an analysis report based on a large model provided by the present invention can customize the content and style of the analysis report according to the actual needs of the enterprise through prompting engineering and report templates, and realize the rapid processing of the large model to generate a report through the method of segmented requests, providing a fast and flexible method for generating an analysis report for the enterprise.

[0090] Figure 3 It is a system structure diagram for generating an enterprise business analysis report based on a preferred embodiment of the present invention.

[0091] As Figure 3 shown, the present invention provides a system for generating an enterprise business analysis report based on a large model, and the system includes:

[0092] An initial unit 301, configured to obtain document data related to the operation of the target unit, parse the document data into markdown format data adapted to the large model; obtain relevant prompts pre-stored in the database;

[0093] Preferably, the initial unit 301 is further configured to: perform information extraction training on the large model through a training data set.

[0094] An assembly unit 302, configured to assemble, based on the prompt, into a question to be asked to the large model by loading the markdown format data;

[0095] A question-asking unit 303, configured to, according to the configured concurrency number, based on the question to be asked to the large model, start multi-threaded question-asking to the large model;

[0096] A generation unit 304, configured to extract information based on the question to be asked to the large model through the large model and generate a question-and-answer result;

[0097] Preferably, the extracted information includes: named entity recognition, relation extraction, and event extraction.

[0098] Preferably, the generation unit 304 is configured to extract information based on questions asked to a large model by the large model to generate a question-and-answer result, and is further configured to:

[0099] Formulate a prompt based on the business scenario, and convert the extracted information into Json data for generating a chart based on the formulated prompt.

[0100] The result unit 305 is configured to fill the question-and-answer result into the business analysis report template file of the target unit to generate a business analysis report in a preset format.

[0101] Preferably, the result unit 305 is configured to fill the question-and-answer result into the business analysis report template file of the target unit to generate a business analysis report in a preset format, and is further configured to:

[0102] Split the business analysis report to obtain the segmented reports after splitting;

[0103] Configure the number of concurrent threads based on the segmented reports after splitting and the computing power information;

[0104] Output the segmented reports based on the configured multiple threads requesting simultaneously.

[0105] Preferably, the result unit 305 is configured to fill the question-and-answer result into the business analysis report template file of the target unit to generate a business analysis report in a preset format, and is further configured to:

[0106] Pre-set a word template, where the word template arranges the style and basic content of the template in advance based on the variable placeholder method, and quickly generates a customized business analysis report through the variable replacement method;

[0107] Convert the business analysis report in the word template into a pdf version and provide it for download by the user side.

[0108] A system for generating a business analysis report of an enterprise based on a large model in a preferred embodiment of the present invention corresponds to a method for generating a business analysis report of an enterprise based on a large model in another preferred embodiment of the present invention, and will not be elaborated herein.

[0109] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) that contain computer-usable program code. The solutions in the embodiments of the present invention can be implemented in various computer languages. For example, object-oriented programming languages such as Java and interpreted scripting languages such as JavaScript, etc.

[0110] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0111] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing devices to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0112] These computer program instructions can also be loaded onto a computer or other programmable data processing devices, such that a series of operation steps are executed on the computer or other programmable devices to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable devices provide steps for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0113] Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications once they learn the basic creative concepts. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications falling within the scope of the present invention.

[0114] Obviously, those skilled in the art can make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

[0115] The present invention has been described by reference to a few embodiments. However, as is well known to those skilled in the art, other embodiments equivalent to those disclosed above of the present invention equally fall within the scope of the present invention as defined by the appended patent claims.

[0116] Generally, all terms used in the claims are to be interpreted according to their ordinary meaning in the technical field, unless otherwise explicitly defined therein. All references to "a / the [device, component, etc.]" are to be interpreted openly as at least one instance of the device, component, etc., unless otherwise explicitly stated. The steps of any method disclosed herein need not be performed in the exact order disclosed, unless explicitly stated.

Claims

1. A method for generating an enterprise operation analysis report based on a large model, the method comprising: Obtain document data related to the target unit's operations, and parse the document data into markdown format data suitable for the large model; Get the relevant prompt stored in the database; Based on the prompt, by loading the markdown format data, assembling the data into questions to be asked to the large model; According to the configured concurrency number, based on the questions raised by the large model, multi-threaded questions are started to be asked to the large model; Extracting information based on the question posed to the big model through the big model to generate a question-answering result; Fill the question and answer results into the business analysis report template file of the target unit to generate a business analysis report in a preset format.

2. The method according to claim 1, further comprising: The large model is trained for information extraction using a training data set.

3. The method according to claim 1, wherein extracting information comprises: Named entity recognition, relation extraction, and event extraction.

4. The method according to claim 1, wherein extracting information based on the question posed to the big model through the big model to generate a question-answering result comprises: Create prompts based on business scenarios, and convert the extracted information into Json data for generating charts based on the created prompts.

5. The method according to claim 1, after filling the question and answer results into the business analysis report template file of the target unit to generate a business analysis report in a preset format, comprises: Split the business analysis report to obtain segmented reports; Configure the number of concurrent threads based on the segmented reports and computing power information after the split; Multiple threads based on the configuration simultaneously request the output of segment reports.

6. The method according to claim 1, wherein the step of filling the question and answer results into the target unit's business analysis report template file to generate a business analysis report in a preset format further comprises: Preset a word template, which pre-arranges the style and basic content of the template based on variable placeholders, and quickly generates a customized business analysis report by variable replacement; Convert the business analysis report in the word template into a pdf version and provide it to the user for download.

7. A system for generating an enterprise operation analysis report based on a large model, the system comprising: The initial unit is used to obtain document data related to the operation of the target unit and parse the document data into markdown format data suitable for the large model; Get the relevant prompt stored in the database; An assembling unit, configured to assemble the data into questions to be asked to the large model based on the prompt by loading the markdown format data; A questioning unit, used for starting multi-threaded questions to the large model based on the questions raised by the large model according to the configured concurrency number; A generating unit, configured to extract information based on the question posed to the large model through the large model, and generate a question-answering result; The result unit is used to fill the question and answer results into the business analysis report template file of the target unit to generate a business analysis report in a preset format.

8. According to the system of claim 7, the initial unit is also used to: perform information extraction training on the large model through a training data set.

9. The system according to claim 7, wherein extracting information comprises: Named entity recognition, relation extraction, and event extraction.

10. The system according to claim 7, wherein the generating unit is used to extract information based on the question posed to the large model through the large model to generate a question-answering result, and is also used to: Create prompts based on business scenarios, and convert the extracted information into Json data for generating charts based on the created prompts.

11. The system according to claim 7, wherein the result unit is used to fill the question and answer result into the business analysis report template file of the target unit to generate a business analysis report in a preset format, and is also used to: Split the business analysis report to obtain segmented reports; Configure the number of concurrent threads based on the segmented reports and computing power information after the split; Multiple threads based on the configuration simultaneously request the output of segment reports.

12. The system according to claim 7, wherein the result unit is used to fill the question and answer results into the business analysis report template file of the target unit to generate a business analysis report in a preset format, and is also used to: Preset a word template, which pre-arranges the style and basic content of the template based on variable placeholders, and quickly generates a customized business analysis report by variable replacement; Convert the business analysis report in the word template into a pdf version and provide it to the user for download.

Images