General report generation method and device, product, equipment and storage medium

Abstract

The invention discloses a general report generation method and device, a product, equipment and a storage medium, and relates to the technical field of report generation, and the method comprises the following steps: carrying out information extraction on mails and chat records related to a preset item through a large language model technology to obtain initial key item information data; formatting the initial key project information data based on a first preset template, and cleaning the formatted initial key project information data to obtain key project information data to be used; and obtaining a general report based on the to-be-used key item information data and a second preset template. By means of the report generation method and device, the technical problem that in the prior art, report generation efficiency and accuracy are low is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of report generation, and in particular, to a general report generation method, device, product, equipment, and storage medium. Background Art

[0002] With the complication of enterprise management requirements, the generation and maintenance of daily reports have become cumbersome and time-consuming, especially the production of regular reports such as bi-weekly reports, weekly reports, and monthly reports.

[0003] Traditional report generation methods usually rely on manual input and collation, which are not only time-consuming and laborious but also prone to data omission and errors. With the expansion of project scale and the diversification of data sources, how to improve the efficiency and accuracy of report generation has become an urgent problem to be solved. Summary of the Invention

[0004] The present invention provides a general report generation method, device, product, equipment, and storage medium, which can solve the technical problem of low efficiency and accuracy in report generation existing in the prior art.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] In a first aspect, an embodiment of the present invention provides a general report generation method, the method comprising:

[0007] Extracting information from emails and chat records related to a preset project through large language model technology to obtain initial key project information data;

[0008] Formatting the initial key project information data based on a first preset template, and cleaning the formatted initial key project information data to obtain key project information data to be used;

[0009] Obtaining a general report based on the key project information data to be used and a second preset template.

[0010] Optionally, the extracting information from emails and chat records related to a preset project through large language model technology to obtain initial key project information data includes:

[0011] Obtaining emails and chat records related to a preset project through robotic process automation technology;

[0012] Parsing the emails and the chat records through optical character recognition technology or large language model technology to obtain the core fields of the emails and the chat records;

[0013] Structurally extracting key information according to the context of the core fields by using large language model technology to obtain initial key project information data.

[0014] Optionally, obtaining a general report based on the key project information data to be used and a second preset template includes:

[0015] Identify the date corresponding to the key project information data to be used, and obtain the date of this identification;

[0016] Calculate the difference between the date of this identification and the date of the last identification, and determine the type of the general report to be generated based on the difference; wherein the type of the general report includes a biweekly report, a weekly report, a cumulative monthly report or a monthly report, and the type of the general report corresponds to the second preset template one by one;

[0017] Based on the type of the general report, selecting a corresponding target preset template from the second preset template;

[0018] The key project information data to be used is filled into the target preset template through the robotic process automation technology to obtain a general report.

[0019] Optionally, after obtaining a general report based on the key project information data to be used and the second preset template, the method further includes:

[0020] If the type of the general report is a biweekly report or a weekly report, the generated general report is obtained once every preset time interval;

[0021] Based on the acquired general report, a general report of the type of cumulative monthly report and / or monthly report is generated through robotic process automation technology and a second preset template.

[0022] Optionally, after obtaining a general report based on the key project information data to be used and the second preset template, the method further includes:

[0023] Using the retrieval enhancement generation model, searching the database for N benchmarking general reports with the highest similarity to the general report, where N is a positive integer;

[0024] Based on N benchmarking general reports, the general reports are tested by using a large language model technology to obtain a test result;

[0025] The second preset template is adjusted based on the detection result to obtain an adjusted second preset template.

[0026] Optionally, the method further comprises:

[0027] Use robotic process automation technology to extract the data required by users from the generated general reports;

[0028] Filter and verify the extracted data through large language model technology, generate a private report based on the filtered and verified data, and send the private report to a preset terminal.

[0029] In a second aspect, an embodiment of the present invention provides a general report generation device, and the device includes:

[0030] An information extraction module, configured to extract information from emails and chat records related to preset items through large language model technology to obtain initial key item information data;

[0031] A data cleaning module, configured to format the initial key item information data based on a first preset template and clean the formatted initial key item information data to obtain key item information data to be used;

[0032] A report generation module, configured to obtain a general report based on the key item information data to be used and a second preset template.

[0033] Optionally, the information extraction module is configured to:

[0034] Obtain emails and chat records related to preset items through robotic process automation technology;

[0035] Parse the emails and the chat records through optical character recognition technology or large language model technology to obtain the core fields of the emails and the chat records;

[0036] Use large language model technology to structurally extract key information according to the context of the core fields to obtain initial key item information data.

[0037] Optionally, the report generation module is configured to:

[0038] Identify the date corresponding to the key item information data to be used to obtain the date of this identification;

[0039] Calculate the difference between the date of this identification and the date of the last identification, and determine the type of the general report to be generated based on the difference; wherein, the types of the general report include bi-weekly report, weekly report, cumulative monthly report or monthly report, and the types of the general report correspond one-to-one with the second preset template;

[0040] Select a corresponding target preset template from the second preset template based on the type of the general report;

[0041] Fill the key item information data to be used into the target preset template through robotic process automation technology to obtain a general report.

[0042] In a third aspect, an embodiment of the present invention further provides an electronic device, including: a memory and a processor; the processor is configured to read and execute a computer program stored in the memory to implement the steps of the foregoing general report generation method.

[0043] In a fourth aspect, an embodiment of the present invention further provides a computer storage medium, in which computer-executable instructions are stored, and when the computer-executable instructions are executed, the steps of the foregoing general report generation method are implemented.

[0044] In a fifth aspect, an embodiment of the present invention further provides a computer program product, including a computer program / instructions, and when the computer program / instructions are executed by a processor, the steps of the foregoing general report generation method are implemented.

[0045] The beneficial effects brought by the technical solutions provided by the embodiments of the present invention include:

[0046] 1. Automated data collection and processing reduce the need for manual operations and improve work efficiency;

[0047] 2. Automatically perform data cleaning and formatting processing, reduce human errors, and improve the accuracy of data processing;

[0048] 3. Combine RPA technology with large speech models to automatically extract data from unstructured data sources (such as emails and chat records), and perform intelligent processing and formatting. RPA technology realizes automated data scraping and input, and the large language model technology is responsible for in-depth semantic analysis and content generation. The combination of the two greatly improves the depth and breadth of data processing, making the content of the generated general report more accurate and rich, and capable of accurately matching project and business requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0050] Figure 1 It is a schematic flowchart of an embodiment of the general report generation method of the present invention;

[0051] Figure 2 It is a schematic diagram of the functional modules of an embodiment of the general report generation device of the present invention;

[0052] Figure 3 It is a schematic structural diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0053] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0054] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings.

[0055] In a first aspect, an embodiment of the present invention provides a general report generation method.

[0056] In one embodiment, referring to Figure 1 , Figure 1 is a flowchart of an embodiment of the general report generation method of the present invention. As Figure 1 shown, the general report generation method includes:

[0057] Step S10, extracting information from emails and chat records related to a preset project through large language model technology to obtain initial key project information data;

[0058] In some specific embodiments, step S10 includes:

[0059] Obtaining emails and chat records related to a preset project through robotic process automation technology;

[0060] Parsing the emails and the chat records through optical character recognition technology or large language model technology to obtain the core fields of the emails and the chat records;

[0061] Using large language model technology, structuring and extracting key information according to the context of the core fields to obtain initial key project information data.

[0062] In this embodiment, a method combining robotic process automation technology and large language model is used to automatically collect progress reports and research data related to a preset project in real time from multiple channels such as chat records and emails.

[0063] Regarding obtaining emails and chat records related to a preset project through robotic process automation technology. Exemplarily, through robotic process automation technology, emails with keywords such as "Xxx Project - xx / xx / xxxx - XX Report" in the email title are automatically identified, and their body or attachments are stored in a specified directory; if the attachment is a PDF, Word, or Excel file, an OCR or conversion tool can be called to extract text and obtain the core fields. Among them, OCR (Optical Character Recognition) refers to the process by which an electronic device (such as a scanner or digital camera) examines printed characters on paper, determines their shape by detecting dark and bright patterns, and then translates the shape into computer text using character recognition methods.

[0064] Chat record collection: For chat software such as QQ, WeCom, and DingTalk, robotic process automation technology simulates the mouse and keyboard, locates and copies or takes screenshots according to a fixed first-line format, such as "Xxx Project - xx / xx / xxxx - Progress Report", and then parses out the core fields through OCR or large language model technology.

[0065] Using large language model technology, according to the context of the core fields of emails and chat records, key information is structurally extracted to obtain initial key project information data. Specifically, the original text or attachment content obtained by robotic process automation technology is sent to the large language model for structural extraction. Among them, context-structured extraction of key information refers to extracting organized and regular information from a piece of text and converting it into a structured format that is easy to analyze and process. This technology is widely used in fields such as data analysis, machine learning, and information retrieval, and can significantly improve data processing efficiency and accuracy.

[0066] For example, the following example can be used:

[0067] "You are a report data parsing assistant. Given the following chat or email content, please extract the following fields:

[0068] 1. Project name (project_name); 2. Report date (report_date); 3. This week's work (this_week_work); 4. Next week's plan (next_week_plan).

[0069] If the field cannot be recognized, set it to an empty string".

[0070] And output the extracted fields in JSON format. After the model outputs, the system parses the JSON format content and maps it to a preliminary data object (such as form: {}). If some fields are empty or cannot be recognized, it will prompt for manual intervention or leave them as empty fields.

[0071] Among them, JSON (JavaScript Object Notation) is designed based on a subset of ECMAScript. It is an open-standard file format and data interchange format that is easy for humans to read and write, and also easy for machines to parse and generate.

[0072] Furthermore, in this embodiment, the data obtained from each channel is also archived according to date, project name, or other tags, which facilitates the aggregation of multiple reported data for the same project later.

[0073] Through this embodiment, the automated information collection, real-time data update and feedback mechanism significantly improves efficiency and accuracy compared with traditional manual input or regular email collection systems, and reduces errors and omissions introduced by human operations. The application of the large language model technology not only improves the quality of data extraction, but also can capture more subtle information differences and optimize the presentation of content, which is superior to traditional extraction models.

[0074] Step S20: Format the initial critical project information data based on a first preset template, and clean the formatted initial critical project information data to obtain the critical project information data to be used;

[0075] In this embodiment, according to the first preset template, the initial critical project information data is formatted through the large language model technology. The first preset template is shown in Table 1.

[0076] Table 1

[0077]

[0078] After obtaining the formatted initial critical project information data, the formatted initial critical project information data is cleaned through robotic process automation technology to obtain the critical project information data to be used. The robotic process automation technology is used in combination with the large language model to conduct audits such as centralized management, permission verification, field integrity check, and comparison with historical records on the captured multi-source data to ensure the authenticity and integrity of the data. For example, for completely duplicate emails or chat records, they can be marked as "duplicate" by robotic process automation technology or scripts to avoid multiple processing. For records with severely abnormal formats (such as all keyword fields being empty), they enter the "to be verified" state and wait for manual or supplementary data. It is easy to understand that data cleaning refers to the process of preprocessing and transforming data to meet the requirements of data analysis and data mining and improve data quality. Data cleaning usually includes operations such as deleting duplicate information, correcting errors, resolving missing values, and deleting outliers from the original data to make subsequent data analysis and data mining more accurate and effective.

[0079] The goal of data cleaning is to improve the accuracy, consistency, and integrity of data, ensuring that the data can truly reflect the business situation and providing a reliable basis for data analysis and decision-making.

[0080] Specifically, first, information such as project leaders, optional report fields, and data collection frequencies is configured in the system. The robotic process automation technology can automatically fill in the corresponding project and personnel information on the configuration platform, reducing manual entry. Different projects may require different fields, such as budgets, risk points, and KPIs. The system can configure form structures for different projects in the database.

[0081] Then, permission verification is carried out: only project leaders or authorized personnel are allowed to modify or submit the data of the project; if the data source does not match, it is marked as "pending review".

[0082] Data quality inspection: The system presets rules, such as blank field inspection, format verification, and logical association. Once potential errors are found, such as incorrect date formats or 100% repetition of fields with historical progress, it will automatically prompt or send a message to the submitter for secondary confirmation.

[0083] Historical comparison: Compare the data of the same project at different times. If it is found to be "exactly the same" as the previous period, the large language model can be called for logical detection and an objection prompt can be given.

[0084] Analyze the text through natural language processing technology (Natural Language Processing, NLP) to extract key information, while removing invalid tone words and correcting logical errors.

[0085] Automatically correct expressions in the text that may cause misunderstandings or deviations, improving the content accuracy and professionalism of the report.

[0086] If a new field (such as value4) is identified in the previous step and has not been configured in the database, the field can be automatically extended; if a required item is missing, an error message is returned to the user or administrator. The error message template is shown in Table 2.

[0087] Table 2

[0088]

[0089] After the cleaning of the formatted initial key project information data is completed, the key project information data to be used can be obtained, and the key project information data to be used is automatically saved in a predetermined data storage system to support long-term tracking and analysis of the data.

[0090] Through this embodiment, before the report data is actually stored in the database, the dual mechanisms of robotic process automation technology and large language model technology are used to check the data accuracy and consistency, greatly reducing the burden of subsequent manual verification and modification. Moreover, by using large language model technology to perform "semantic cleaning" and "logical error correction" on the text and then writing it into the database, a large amount of repetitive labor at the formatting, polishing, or checking levels can be eliminated, making the quality of the subsequent generated general report better and making statistical analysis or historical comparison easier.

[0091] Step S30: Obtain a general report based on the key project information data to be used and the second preset template.

[0092] In some specific embodiments, step S30 includes:

[0093] Identify the date corresponding to the key project information data to be used to obtain the date identified this time;

[0094] Calculate the difference between the date identified this time and the date identified last time, and determine the type of the general report to be generated based on the difference; wherein, the types of the general report include bi-weekly report, weekly report, cumulative monthly report, or monthly report, and the types of the general report correspond one-to-one with the second preset template;

[0095] Select the corresponding target preset template from the second preset template based on the type of the general report;

[0096] Fill the key project information data to be used into the target preset template through robotic process automation technology to obtain a general report.

[0097] In this embodiment, the key project information data to be used is structured and converted into structured report content. The core functions of this step include automatically identifying the current date of the key project information data to be used, and at the same time applying preset templates and formats to organize the information, and accordingly determining whether to generate a bi-weekly report, weekly report, cumulative monthly report, or monthly report.

[0098] Specifically, identify the date corresponding to the key project information data to be used to obtain the date identified this time; calculate the difference between the date identified this time and the date identified last time.

[0099] Determine the type of the general report to be generated based on the difference. The types of the general report include weekly report, bi-weekly report, monthly report, or cumulative monthly report. Specifically, weekly report: generated based on the data of the current week; bi-weekly report: generated based on the data of two consecutive weeks; monthly report: generated based on the data of the current month; cumulative monthly report: generated based on the cumulative monthly report data of the current month and the previous month, that is, the monthly report of each month is cumulative, and adding the cumulative monthly report of the previous month to the monthly report of the current month can obtain a new cumulative monthly report.

[0100] In this embodiment, the types of general reports correspond one-to-one with the second preset templates. After determining the type of the general report, select the target preset template corresponding to the type of the general report from the second preset templates, and fill the key project information data to be used into the target preset template through robotic process automation technology, then the general report can be obtained. Exemplarily, the target preset template corresponding to a weekly report is shown in Table 3.

[0101] Table 3

[0102]

[0103] Furthermore, record the generated general report into the corresponding fields (such as this_week_work, weekly_summary, etc.) to improve the readability of the report. At the same time, store the corresponding data into database tables such as reports_daily, reports_weekly, reports_month, etc., or distinguish the periods by fields in the same table. If the project requirements change, the robotic process automation technology can automate the operation system configuration interface again to add new second preset templates or fields. If the large language model technology discovers semantic conflicts during text merging or rectification, such as self-contradictions in the same paragraph, it can mark the errors during output for further inspection or prompt manual intervention for modification later.

[0104] The robotic process automation technology realizes the automated extraction and input of data, and the large language model technology is responsible for in-depth semantic analysis and content generation. The combination of the two greatly improves the depth and breadth of data processing.

[0105] In this embodiment, information extraction is performed on emails and chat records related to preset projects through the large language model technology to obtain initial key project information data; format the initial key project information data based on the first preset template, and clean the formatted initial key project information data to obtain the key project information data to be used; obtain a general report based on the key project information data to be used and the second preset template. Through this embodiment, the automated collection and processing of data reduce the need for manual operations, improve work efficiency, automatically perform data cleaning and formatting processing, reduce human errors, improve the accuracy of data processing, and finally obtain a general report based on the processed data and the second preset template, improving the accuracy of the generated general report and solving the technical problems of low efficiency and accuracy in report generation in the prior art.

[0106] Optionally, in one embodiment, after obtaining the general report based on the key project information data to be used and the second preset template, it includes:

[0107] If the type of the general report is a weekly report or a bi - weekly report, obtain the generated general report once every preset time interval.

[0108] Based on the obtained general report, generate general reports of the cumulative monthly report type and / or monthly report type through robotic process automation technology and a second preset template.

[0109] In this embodiment, if the general report generated in step S30 is a weekly report or a bi - weekly report, accumulate the bi - weekly reports or weekly reports to generate a cumulative monthly report and / or monthly report. Specifically, every preset time interval, use robotic process automation technology to obtain the generated general report once, automatically log in to the enterprise management platform or financial system, fill in the corresponding fields on the corresponding input page, and then submit and save. For the general reports that need to be exported, the RPA technology automatically applies data according to the corresponding second preset template and generates general reports of the cumulative monthly report type and / or monthly report type.

[0110] Furthermore, integrate a large model into the RPA process to process unstructured or semi - structured data, such as emails or chat records. The large model provides decision - making support to help automatically judge the compliance of the data and whether manual review is required.

[0111] The large language model technology can be used again for "title generation", "abstract addition", "rhetorical polishing" or "logical verification". For example: "This is the content of the weekly report generated by the system:... Please supplement the brief project highlights based on the existing content and summarize it in one paragraph."

[0112] According to feedback or permissions, the large language model technology can also be used to filter sensitive information to ensure the security of the general report distribution. The general report distribution includes multi - channel automatic distribution and individual distribution.

[0113] Multi - channel automatic distribution: The RPA technology can automatically send the generated report files or links to the relevant responsible person's email, instant messaging group or store them in the shared document system, without manual uploading one by one.

[0114] Individual distribution to external partners: Additional format or field adjustments can be made to the general report.

[0115] Through this embodiment, "automatic form filling + model - assisted text optimization + multi - channel distribution" is integrated, greatly reducing repetitive operations. When the number of projects is large or the frequency is high, the efficiency advantage of the RPA technology can be more reflected. Among them, RPA (Robotic Process Automation) is a technology that realizes business process automation by software robots simulating human operations on a computer. It can automatically execute repetitive and regular tasks, such as data entry, file processing, and report generation.

[0116] Optionally, in one embodiment, after obtaining the general report based on the key project information data to be used and the second preset template, the following steps are included:

[0117] Use a retrieval-augmented generation model to retrieve the top N benchmark general reports with the highest similarity to the general report in the database, where N is a positive integer;

[0118] Based on the N benchmark general reports, use large language model technology to detect the general report and obtain the detection result;

[0119] Adjust the second preset template based on the detection result to obtain the adjusted second preset template.

[0120] In this embodiment, a retrieval-augmented generation model (Retrieval-augmented Generation, RAG) is used to learn from historical data and user feedback, and automatically adjust algorithm parameters according to past mistakes and successful experiences to improve the effect of report summary generation to better meet the requirements.

[0121] Specifically, whenever a new general report is generated, use the retrieval-augmented generation model to retrieve the top N benchmark general reports with the highest similarity to the generated general report in the database, where N is a positive integer.

[0122] Based on the N benchmark general reports, use large language model technology to detect the current general report and make logical judgments in a richer context.

[0123] For example: "This is a weekly report summary of Project X in the past three weeks:

[0124] Week 1:...;

[0125] Week 2:...;

[0126] Week 3:...;

[0127] Please generate a report for this week that is reasonably connected to the previous progress, with a word count of approximately 200 words."

[0128] If the user often modifies the generated text in the general report, the system will record "before / after modification", analyze the reasons for errors, such as non-compliance with format requirements, inaccurate translation of professional terms, etc., and improve the general report generated next time by fine-tuning the program or increasing RAG index materials. And within a certain period (such as monthly), a model quality assessment will be carried out, statistics will be made on the fields / sentence patterns with a high manual correction rate in the general report, the extraction and generation strategies will be automatically updated, and the second preset template will be adjusted based on the detection result to obtain the adjusted second preset template, forming a positive feedback loop to continuously improve the automation level.

[0129] Optionally, in one embodiment, the method further includes:

[0130] Extracting the data required by the user from the generated general report using robotic process automation technology;

[0131] Filtering and validating the extracted data through large language model technology, generating a private report based on the filtered and validated data, and sending the private report to a preset terminal.

[0132] In this embodiment, according to specific requirements, key information is extracted from the general report. Based on the same general report, "privatized streamlining" is performed for different roles or requirements, automatically excluding or hiding certain fields that are not suitable for public disclosure, generating a private report, and comparing the project names through a large language model. Using RPA to automatically complete data extraction, report generation, and sending to the designated recipient.

[0133] Specifically, use RPA technology to extract the data required by the user from the generated general report, and reprocess it through field filtering or large language model technology to retain only specific information and generate a customized private report.

[0134] For example, "The following is the text of this week's comprehensive report:

[0135] Xxxxxxxx

[0136] Please remove the finance and human resources fields, and only retain the project progress, problems to be solved, and next week's plan, and output a streamlined private report."

[0137] Apply large language model technology to filter and validate the data required by the user to ensure the accuracy of the data. By optimizing the report content, the generated private report better meets the specific needs and format requirements of the user.

[0138] When there are multiple projects, the model can be made to only screen paragraphs related to a certain project, or a "role permission" field can be added to the database to allow the system to automatically exclude confidential fields when generating a private report.

[0139] After generating the private report, automatically send the generated private report to the designated recipient, such as a corporate WeChat group, email list, etc. Ensure the timely delivery of the report and the accuracy of receipt. Exemplarily, use RPA technology to send the private version report to the designated team or external unit, such as an email list, corporate WeChat, DingTalk group, etc.;

[0140] If multiple versions of the report are required (such as a finance version, a public version, a leadership-exclusive version), they can be batch-generated with one click and automatically delivered.

[0141] Through this embodiment, the multiplexing effect of collecting and processing data once and outputting differentially multiple times is realized, which not only protects sensitive information but also meets the streamlined requirements of reports for different roles, taking into account both efficiency and security.

[0142] In a second aspect, an embodiment of the present invention further provides a general report generation device.

[0143] In one embodiment, referring to Figure 2 , Figure 2 is a schematic diagram of the functional modules of an embodiment of the general report generation device of the present invention. As Figure 2 shown, the general report generation device includes:

[0144] An information extraction module 10, configured to extract information from emails and chat records related to a preset project through large language model technology to obtain initial key project information data;

[0145] A data cleaning module 20, configured to format the initial key project information data based on a first preset template and clean the formatted initial key project information data to obtain key project information data to be used;

[0146] A report generation module 30, configured to obtain a general report based on the key project information data to be used and a second preset template.

[0147] Optionally, in one embodiment, the information extraction module 10 is configured to:

[0148] Obtain emails and chat records related to a preset project through robotic process automation technology;

[0149] Parse the emails and the chat records through optical character recognition technology or large language model technology to obtain the core fields of the emails and the chat records;

[0150] Use large language model technology to structurally extract key information according to the context of the core fields to obtain initial key project information data.

[0151] Optionally, in one embodiment, the report generation module 30 is configured to:

[0152] Identify the date corresponding to the key project information data to be used to obtain the date of this identification;

[0153] Calculate the difference between the date of this identification and the date of the last identification, and determine the type of the general report to be generated based on the difference; wherein, the types of the general report include bi-weekly reports, weekly reports, cumulative monthly reports or monthly reports, and the types of the general report correspond one-to-one with the second preset template;

[0154] Select a corresponding target preset template from the second preset templates based on the type of the general report;

[0155] Fill the key item information data to be used into the target preset template through robotic process automation technology to obtain a general report.

[0156] Optionally, in one embodiment, the report generation module 30 is configured to:

[0157] If the type of the general report is a bi-weekly report or a weekly report, obtain the generated general report once every preset time interval;

[0158] Based on the obtained general report, generate general reports of the cumulative monthly report type and / or monthly report type through robotic process automation technology and the second preset template.

[0159] Optionally, in one embodiment, the device further includes a parameter adjustment module, which is configured to:

[0160] Use a retrieval-augmented generation model to retrieve the top N benchmark general reports with the highest similarity to the general report in the database, where N is a positive integer;

[0161] Based on the N benchmark general reports, detect the general report through large language model technology to obtain a detection result;

[0162] Adjust the second preset template based on the detection result to obtain an adjusted second preset template.

[0163] Optionally, in one embodiment, the report generation module 30 is further configured to:

[0164] Use robotic process automation technology to extract the data required by the user from the generated general report;

[0165] Filter and verify the extracted data through large language model technology, generate a private report based on the filtered and verified data, and send the private report to a preset terminal.

[0166] Wherein, the function implementation of each module in the above general report generation device corresponds to each step in the above general report generation method embodiment, and its function and implementation process will not be elaborated here one by one.

[0167] In a third aspect, an embodiment of the present invention further provides an electronic device, the structure of which is as Figure 3 shown, including: a memory, a processor, and the processor is used to read and execute the computer program stored in the memory to implement the foregoing general report generation method.

[0168] Fourthly, an embodiment of the present invention further provides a computer storage medium, in which computer-executable instructions are stored, and when the computer-executable instructions are executed, the foregoing general report generation method is implemented.

[0169] Fifthly, an embodiment of the present invention provides a computer program product, which is stored in a storage medium and is executed by at least one processor to implement each process of the foregoing general report generation method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be elaborated here.

[0170] Finally, it should be noted that in some processes described in the embodiments of the present invention, multiple operations or steps appear in a specific order. However, it should be understood that these operations or steps may not be executed in the order in which they appear in the embodiments of the present invention or may be executed in parallel. The serial numbers of the operations are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed in sequence or in parallel, and these operations or steps may be combined.

[0171] The foregoing are only the preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or perform equivalent replacements for some of the technical features. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A general report generation method, characterized in that: The method comprises: Use large language model technology to extract information from emails and chat records related to preset projects to obtain initial key project information data; Formatting the initial key project information data based on a first preset template, and cleaning the formatted initial key project information data to obtain the key project information data to be used; A general report is obtained based on the key project information data to be used and the second preset template.

2. The general report generation method according to claim 1, characterized in that: The large language model technology is used to extract information from emails and chat records related to the preset project to obtain initial key project information data, including: Capture emails and chat logs related to pre-set projects through robotic process automation technology; Parsing the email and the chat record by using optical character recognition technology or large language model technology to obtain core fields of the email and the chat record; By using the large language model technology, key information is structured and extracted according to the context of the core fields to obtain initial key project information data.

3. The general report generation method according to claim 1, characterized in that: The method of obtaining a general report based on the key project information data to be used and the second preset template includes: Identify the date corresponding to the key project information data to be used, and obtain the date of this identification; Calculate the difference between the date of this identification and the date of the last identification, and determine the type of the general report to be generated based on the difference; wherein the type of the general report includes a biweekly report, a weekly report, a cumulative monthly report or a monthly report, and the type of the general report corresponds to the second preset template one by one; Based on the type of the general report, selecting a corresponding target preset template from the second preset template; The key project information data to be used is filled into the target preset template through the robotic process automation technology to obtain a general report.

4. The general report generation method according to claim 3, characterized in that: After obtaining a general report based on the key project information data to be used and the second preset template, the method includes: If the type of the general report is a biweekly report or a weekly report, the generated general report is obtained once every preset time interval; Based on the acquired general report, a general report of the type of cumulative monthly report and / or monthly report is generated through robotic process automation technology and a second preset template.

5. The general report generation method according to claim 1, characterized in that: After obtaining a general report based on the key project information data to be used and the second preset template, the method includes: Using the retrieval enhancement generation model, searching the database for N benchmarking general reports with the highest similarity to the general report, where N is a positive integer; Based on N benchmarking general reports, the general reports are tested by using a large language model technology to obtain a test result; The second preset template is adjusted based on the detection result to obtain an adjusted second preset template.

6. The general report generation method according to any one of claims 1 to 5, characterized in that: The method further comprises: Use robotic process automation technology to extract the data required by users from the generated general reports; The extracted data is filtered and verified by using a large language model technology, a private report is generated based on the filtered and verified data, and the private report is sent to a preset terminal.

7. A general report generation device, characterized in that: The device comprises: An information extraction module is configured to extract information from emails and chat records related to a preset project using a large language model technology to obtain initial key project information data; A data cleaning module is configured to format the initial key project information data based on a first preset template, and clean the formatted initial key project information data to obtain the key project information data to be used; The report generation module is configured to obtain a general report based on the key project information data to be used and a second preset template.

8. The general report generation device according to claim 7, characterized in that: The information extraction module is configured to: Capture emails and chat logs related to pre-set projects through robotic process automation technology; Parsing the email and the chat record by using optical character recognition technology or large language model technology to obtain core fields of the email and the chat record; By using the large language model technology, key information is structured and extracted according to the context of the core fields to obtain initial key project information data.

9. The general report generation device according to claim 7, characterized in that: The report generation module is configured to: Identify the date corresponding to the key project information data to be used, and obtain the date of this identification; Calculate the difference between the date of this identification and the date of the last identification, and determine the type of the general report to be generated based on the difference; wherein the type of the general report includes a biweekly report, a weekly report, a cumulative monthly report or a monthly report, and the type of the general report corresponds to the second preset template one by one; Based on the type of the general report, selecting a corresponding target preset template from the second preset template; The key project information data to be used is filled into the target preset template through the robotic process automation technology to obtain a general report.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instruction is executed by a processor, the steps of the general report generation method as described in any one of claims 1 to 6 are implemented.

11. An electronic device, characterized in that: include: Memory and processor; The processor is used to read and execute the computer program stored in the memory to implement the steps of the general report generation method as described in any one of claims 1 to 6.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed, the steps of the general report generation method according to any one of claims 1 to 6 are implemented.

Images