Intelligent project auditing method and system based on multi-agent fusion

Abstract

The invention discloses an intelligent project auditing method and system based on multi-agent fusion, and the method comprises the steps: obtaining a project application text, carrying out the pre-auditing processing, analyzing the content logic relation of the project text after the pre-auditing, and obtaining project text auditing data, extracting text features of the project text review data, analyzing potential problems in the project declaration document in combination with the corresponding content logic relationship to obtain project problem analysis data, and performing multi-dimensional compiling processing on the project text review data and the project problem analysis data according to a preset project review model to obtain project text review data and project problem analysis data; and obtaining a project review analysis report conforming to the corresponding project review model, performing key data verification processing on the project review analysis report, evaluating the whole project declaration compliance according to a key data verification result, and feeding back and adjusting the whole project review process according to a compliance evaluation result. The method has the effects of improving the auditing accuracy of massive complex project declaration materials and improving the project material auditing efficiency.

Description

Technical Field

[0001] This invention relates to the technical field of project review, and in particular to an intelligent project review method and system based on multi-agent fusion. Background Technology

[0002] Currently, as enterprises and institutions place increasing emphasis on science and technology projects, the number of project application materials is also increasing, leading to a surge in the workload of reviewing these materials.

[0003] Current methods for reviewing project application materials typically rely on artificial intelligence to extract information such as titles from application documents. However, detailed project data still requires manual review. Furthermore, each stage of the project application process, from initiation and evaluation to final approval, generates a large volume of complex textual materials. Reviewers need to repeatedly consult relevant documents and compare key information to confirm the accuracy of the application materials. When dealing with a massive amount of application materials from multiple projects, this process is prone to errors and impacts efficiency. Therefore, there is room for further optimization in the aforementioned technologies regarding the review methods for handling large volumes of complex application materials from multiple projects. Summary of the Invention

[0004] To address the problems of errors and low efficiency in the review of massive and complex application materials in existing technologies, this invention provides an intelligent project review method and system based on multi-agent fusion, which can improve the accuracy and efficiency of reviewing massive and complex project application materials.

[0005] Firstly, the above-mentioned inventive objective of this application is achieved through the following technical solution: An intelligent project review method based on multi-agent fusion, the method comprising: Obtain the project application text, conduct a preliminary review of the project application text, and analyze the logical relationship of the content of the pre-reviewed project text to obtain project text review data; Extract text features from the project text review data, and analyze potential problems in the project application documents based on the text features and corresponding content logical relationships to obtain project problem analysis data; Based on the preset project review model, the project text review data and the project problem analysis data are processed in multiple dimensions to obtain a project review analysis report that conforms to the corresponding project review model. The project review and analysis report is subjected to key data verification processing. The compliance of the entire project application is assessed based on the key data verification results, and the project review process is adjusted based on the compliance assessment results.

[0006] In a preferred embodiment, this application can be further configured as follows: obtaining the project application text, performing a preliminary review on the project application text, and analyzing the logical relationships of the content of the pre-reviewed project text to obtain project text review data, specifically including: Obtain the project application text, conduct a preliminary review of the project application text according to the preset text review conditions, and obtain the text pre-review data that meets the preset text review conditions; The text pre-review data is associated with each sub-project of the project, and the logical relationship between the content of each sub-project is analyzed to obtain the logical analysis data between each sub-project. Obtain the phased application data for each application stage of the project, perform consistency analysis of the application materials based on the phased application data and the logical analysis data, and evaluate the trend of differences in application materials at different application stages; Based on the trends in the differences in the application materials, the project texts with differences in the application materials are marked to obtain the project text review data for the project applications.

[0007] In a preferred embodiment, this application can be further configured as follows: extracting text features from the project text review data, and analyzing potential problems in the project application documents based on the text features and corresponding content logical relationships to obtain project problem analysis data, specifically including: The project text review data is processed by extracting key text features, and the project application results are predicted based on the text features and the corresponding content logic relationship to obtain the project application prediction results. Obtain the actual application results for the corresponding projects, and compare the project application prediction results with the actual application results according to the logical relationship of the content, and analyze the accuracy of the project application prediction. Based on the project application prediction accuracy, potential problems in the project application documents are identified and summarized according to the corresponding project application stage to obtain project problem analysis data.

[0008] In a preferred embodiment, this application can be further configured as follows: before performing multi-dimensional writing processing on the project text review data and the project problem analysis data according to a preset project review model to obtain a project review analysis report that meets the corresponding report writing requirements, the application further includes: Obtain application case data of similar projects, use the application case data of similar projects as data samples for data training, and annotate the key text of the current project type to obtain text annotation data; Based on the text annotation data, analyze the text review conditions and model boundaries for the current project type, and construct a project review model.

[0009] In a preferred embodiment, this application can be further configured as follows: Based on a preset project review model, the project text review data and the project problem analysis data undergo multi-dimensional processing to obtain a project review analysis report that meets the corresponding report writing requirements, specifically including: Based on the preset project review model, the project text review data and the project problem analysis data are used to conduct a project application impact analysis, and the quality impact factors of project problems on the project application quality are generated. Extract the corresponding key text data according to the preset report review conditions, predict the project review results of the key text data according to the quality influencing factors, and obtain the project review prediction results corresponding to the current report review conditions. According to the preset report writing requirements, the key text data and the corresponding project review prediction results are processed to obtain a project review analysis report that meets the corresponding report writing requirements.

[0010] In a preferred embodiment, this application can be further configured as follows: performing key data verification processing on the project review and analysis report, assessing the compliance of the entire project application based on the key data verification results, and adjusting the entire project review process based on the compliance assessment results, specifically including: Key data in the project review analysis report are extracted based on preset project review keywords. The key data is then compared with the corresponding project application standard parameters to obtain the key data verification results. Based on the key data verification results, analyze the importance of each key data in the entire project application, assess the compliance of the entire project application based on the importance analysis results, and generate a compliance assessment result; The compliance assessment results are fed back to each project application stage, and the text review conditions for each project application stage are adjusted accordingly, with adaptive optimization processes implemented for each project application stage.

[0011] In a preferred embodiment, this application can be further configured as follows: after performing key data verification processing on the project review and analysis report, assessing the compliance of the entire project application based on the key data verification results, and adjusting the entire project review process based on the compliance assessment results, it also includes: Obtain the review progress of each application stage of the project, conduct a full-process review of the data upload time and corresponding uploaded review text for each stage of the review process, and obtain the review results of each stage of the process. Based on the review results of the aforementioned stages, it is determined whether the project deliverables for the corresponding application stage meet the preset delivery indicators, and project deliverables that do not meet the preset delivery indicators are extracted to generate the corresponding stage review report.

[0012] Secondly, the above-mentioned inventive objective of this application is achieved through the following technical solutions: An intelligent project review system based on multi-agent fusion, the system being applied to the aforementioned intelligent project review method based on multi-agent fusion, the system comprising: The intelligent text review module is used to acquire project application texts, perform pre-review processing on the project application texts, and analyze the content logic relationship of the pre-reviewed project texts to obtain project text review data. The intelligent problem analysis module is used to extract text features from the project text review data, and analyze potential problems in the project application documents based on the text features and the corresponding content logic relationships to obtain project problem analysis data. The intelligent report writing module is used to perform multi-dimensional writing and processing on the project text review data and the project problem analysis data according to the preset project review model, so as to obtain a project review analysis report that conforms to the corresponding project review model. The process automation module is used to perform key data verification processing on the project review and analysis report, evaluate the compliance of the entire project application based on the key data verification results, and adjust the entire project review process based on the compliance assessment results.

[0013] Thirdly, the above-mentioned objectives of this application are achieved through the following technical solutions: A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described intelligent project review method based on multi-agent fusion.

[0014] Fourthly, the above-mentioned objectives of this application are achieved through the following technical solutions: A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described intelligent project review method based on multi-agent fusion.

[0015] In summary, this application includes at least one of the following beneficial technical effects: 1. This application conducts a review and analysis of the entire process of science and technology projects, including a joint analysis of text material characteristics and key review stages. It uses a project review model to conduct text form review, content logic analysis, and correlation analysis, and outputs corresponding review and analysis reports in a multi-dimensional report format. It also provides feedback and adjustment to the entire review process based on the project compliance assessment results. During the feedback process, it continuously adjusts the review conditions of each review stage according to the actual situation, thereby achieving automated and intelligent review of science and technology project documents and improving the accuracy of reviewing massive amounts of complex project application materials. Compared with manual review, this application improves the efficiency of project material review. 2. In this invention, key data is labeled in the project review model to provide high-quality training samples for model training, improve the model's accuracy in learning and understanding project text features and patterns, thereby improving the accuracy and efficiency of the review. Furthermore, by introducing corresponding project review conditions, the model boundary is constructed to improve the accuracy of report preparation. 3. This application improves the timeliness of monitoring the entire project review process and its compatibility with actual delivery by monitoring the progress of each stage of the project application process and analyzing and providing feedback on the upload time and data of project deliverables in a timely manner. Attached Figure Description

[0016] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0017] Figure 1 This is a flowchart illustrating the implementation of an intelligent project review method based on multi-agent fusion in this embodiment.

[0018] Figure 2 This is a flowchart illustrating the implementation of step S10 of the intelligent project review method in this embodiment.

[0019] Figure 3 This is a flowchart illustrating the implementation of step S20 of the intelligent project review method in this embodiment.

[0020] Figure 4 This is a flowchart illustrating the implementation of the intelligent project review method construction model in this embodiment.

[0021] Figure 5 This is a flowchart illustrating the implementation of step S30 of the intelligent project review method in this embodiment.

[0022] Figure 6 This is a flowchart illustrating the implementation of step S40 of the intelligent project review method in this embodiment.

[0023] Figure 7 This is a flowchart illustrating the implementation of the intelligent project review method in this embodiment for phased review.

[0024] Figure 8 This is a structural block diagram of the intelligent project review system based on multi-agent fusion in this embodiment.

[0025] Figure 9 This is a schematic diagram of the internal structure of a computer device used to implement intelligent project review methods. Detailed Implementation

[0026] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0028] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0029] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0030] In one embodiment, such as Figure 1 As shown, this application discloses an intelligent project review method based on multi-agent fusion, which specifically includes the following steps: S10: Obtain the project application text, conduct a preliminary review of the project application text, analyze the logical relationship of the content of the pre-reviewed project text, and obtain project text review data.

[0031] Specifically, such as Figure 2 As shown, step S10 includes: S101: Obtain the project application text, conduct a preliminary review of the project application text according to the preset text review conditions, and obtain the text pre-review data that meets the preset text review conditions.

[0032] Specifically, the project application text is obtained by scanning the project application documents, and the project application text is pre-examined according to preset text review conditions. In this embodiment, the text review conditions include text duplication review, text accuracy review, text consistency review, and text compliance review. The pre-examination data that meets the preset text review conditions is obtained through the pre-examination.

[0033] S102: Associate the pre-reviewed text data with each sub-project of the project, analyze the logical relationship between the content of each sub-project, and obtain the logical analysis data between each sub-project.

[0034] Specifically, according to the extraction location of the project application text and the project scope of each sub-project, the pre-review data of the text is associated with each sub-project of the project, and the logical relationship between the content of each sub-project is analyzed. Specifically, the logical relationship between the content is analyzed through the research content, technical route and expected results of the project, so as to obtain the logical analysis data of each sub-project.

[0035] S103: Obtain the phased application data for each application stage of the project, conduct consistency analysis of the application materials based on the phased application data and logical analysis data, and assess the trend of differences in application materials at different application stages.

[0036] Specifically, according to the different application stages before, during, and after project approval, phased application data for each stage is obtained. Based on the phased application data and logical analysis data, a consistency analysis of the application materials is conducted to ensure logical consistency between different application stages. The consistency of application materials includes consistency of research content, technical route, and expected results. Research content includes the consistency of data such as terminology, definitions, formulas, and charts across different application stages. Based on the consistency analysis results, the trend of differences in application materials across different application stages is assessed. For example, based on the research route and expected results, the trend of changes in application materials in adjacent application stages is predicted and compared with the materials in the current consistency assessment results. Based on the comparison results, the application stages with material differences are located, and the trend of differences in application materials in the corresponding application stages is determined.

[0037] S104: Based on the trend of differences in the application materials, the project texts with differences in the application materials are marked to obtain the project text review data of the project application.

[0038] Specifically, based on the trends in discrepancies among the submitted materials, project texts with discrepancies are annotated, including annotations for completeness, consistency, compliance, semantic errors, budget errors, target achievement, and duplicate submissions. This text annotation yields project application text review data. The purpose of this annotation is to provide high-quality training data for the automated science and technology project text review algorithm model, enabling it to better learn and understand the characteristics and patterns of the text, thereby improving the accuracy and efficiency of the review process.

[0039] S20: Extract text features from the project text review data, and analyze potential problems in the project application documents based on the text features and corresponding content logic relationships to obtain project problem analysis data.

[0040] Specifically, such as Figure 3 As shown, step S20 includes: S201: Extract key text features from the project text review data, and predict the project application results based on the text features and the corresponding content logic relationships to obtain the project application prediction results.

[0041] Specifically, key text features are extracted from the project text review data based on the key indicators of project review, such as the parameters and quotation amount corresponding to the key indicators. Based on the text features and the corresponding content logic relationship, such as the research logic of the technical route, the project application results are predicted. If the key text features meet the content logic relationship and the requirements of the project application indicators, the project application is predicted to meet the application requirements.

[0042] S202: Obtain the actual application results for the corresponding projects, compare the project application prediction results with the actual application results based on the logical relationship of the content, and analyze the accuracy of the project application prediction.

[0043] Specifically, the actual application results of projects at different application stages are obtained. According to the logical relationship of the content, the predicted application results and the actual application results are compared with key text features to obtain the key text differences between the predicted and actual application results, including project name, research content, expected results, budget, etc. The accuracy of project application prediction is analyzed based on the ratio of the difference key text features to the total key text features. The larger the ratio, the lower the accuracy of project application prediction.

[0044] S203: Based on the accuracy of project application prediction, identify potential problems in project application documents and summarize them according to the corresponding project application stage to obtain project problem analysis data.

[0045] Specifically, based on the accuracy of project application predictions, the difference trends of key texts are analyzed. Combined with the corresponding content logic relationships, all key texts related to the difference trends and content logic are summarized to generate potential problems in the project application documents. According to the corresponding project application stage, all potential problems are summarized to obtain project problem analysis data.

[0046] In this embodiment, the model's ability to capture problematic documents is also evaluated by comparing the potential problems identified by the model with the total number of problematic documents actually annotated by humans.

[0047] S30: Based on the preset project review model, process the project text review data and project problem analysis data in multiple dimensions to obtain a project review analysis report that conforms to the corresponding project review model.

[0048] Specifically, before processing the project text review data and project issue analysis data in multiple dimensions according to the preset project review model to obtain a project review analysis report that meets the corresponding report writing requirements, such as... Figure 4 As shown, it also includes: S301: Obtain case data of similar project applications, use the case data of similar project applications as data samples for data training, and annotate the key text of the current project type to obtain text annotation data.

[0049] Specifically, data on similar project application cases are obtained according to the current project application type. The data on similar project application cases is trained using BERT or XLNet big data algorithms. Based on the training results, the key text information of the current project type is obtained and labeled to obtain text-annotated data.

[0050] S302: Based on the text annotation data, analyze the text review conditions and model boundaries of the current project type, and construct a project review model.

[0051] Specifically, based on the text annotation data, the text review conditions and model boundaries of the current project type are analyzed, including project feasibility, project economics and financial compliance as model boundary conditions. The text review conditions mainly include goal completion, process management, output, technical level, talent training, impact of results, economic benefits and social benefits, thereby constructing a project review model.

[0052] Specifically, such as Figure 5 As shown, step S30 specifically includes: S303: Based on the preset project review model, conduct project application impact analysis on project text review data and project problem analysis data, and generate quality impact factors of project problems on project application quality.

[0053] Specifically, the project text review data and project problem analysis data are input into a preset project review model to conduct project application impact analysis. Based on the importance of potential problems in the project problem analysis data to the project application, the impact of potential problems on the project application is assessed. In conjunction with the project text review data of potential problems, the most influential key parameters such as quotation amount and signature and seal are analyzed to determine the quality impact factors of project problems on the quality of project application.

[0054] S304: Extract the corresponding key text data according to the preset report review conditions, predict the project review results based on the key text data according to the quality influencing factors, and obtain the project review prediction results corresponding to the current report review conditions.

[0055] Specifically, based on preset report review conditions such as technical level, talent cultivation, impact of achievements, economic benefits and social benefits, key text data corresponding to the report review conditions are extracted to construct an evaluation matrix. Local science and technology innovation data and relevant domestic and foreign data are used as indicators, and quality influencing factors are introduced as constraints to establish evaluation standards. The project review results are predicted based on the key text data, thereby obtaining the project review prediction results corresponding to the current report review conditions.

[0056] S305: In accordance with the preset report writing requirements, process the key text data and the corresponding project review prediction results to obtain a project review analysis report that meets the corresponding report writing requirements.

[0057] Specifically, the model's report writing conditions are adjusted according to different report requirements. Key text data and corresponding project review prediction results are extracted, and materials are written according to preset report writing conditions to obtain a project review analysis report that meets the corresponding report writing requirements. In this embodiment, multi-dimensional project review analysis reports can be generated by adjusting different writing requirements, thereby improving the intelligence level of project review analysis report writing.

[0058] S40: Perform key data verification processing on the project review and analysis report, assess the compliance of the entire project application based on the key data verification results, and adjust the entire project review process based on the compliance assessment results.

[0059] Specifically, such as Figure 6 As shown, step S40 includes: S401: Extract key data from the project review analysis report based on preset project review keywords, compare the key data with the corresponding project application standard parameters, and obtain the key data verification results.

[0060] Specifically, key data in the project review analysis report are extracted according to preset project review keywords. The key data is then compared with the corresponding project application standard parameters, such as whether the preset positions of key indicators like stamps and signatures are blank, thereby obtaining the key data verification results.

[0061] S402: Based on the key data verification results, analyze the importance of each key data in the entire project application, assess the compliance of the entire project application based on the importance analysis results, and generate a compliance assessment result.

[0062] Specifically, based on the verification results of key data, the importance of each key data in the entire project application is analyzed. For example, if there is no stamp or signature, the application cannot be submitted, indicating high importance. For example, formal errors such as grammatical errors, spelling errors, inappropriate wording, and punctuation errors can be corrected later and have little impact on the project application approval, indicating low importance. Thus, a corresponding importance coefficient is assigned to each key data. In conjunction with whether the key data complies with preset laws, regulations, government provisions, and industry standards, the compliance of the entire project application is assessed, and a compliance assessment result is generated.

[0063] S403: Feedback the compliance assessment results to each project application stage, adjust the text review conditions for each project application stage, and perform adaptive optimization for each project application stage.

[0064] Specifically, the compliance assessment results will be fed back to each project application stage, and the text review conditions for each project application stage will be adjusted accordingly. For example, if there is non-compliant key data, the application process will be traced back to its source, and the text review conditions for the entire application process will be adjusted until the key data meets the compliance requirements. The review conditions for each project application stage will be adaptively optimized.

[0065] In this embodiment, as Figure 7 As shown, after performing key data verification on the project review and analysis report, assessing the compliance of the entire project application based on the key data verification results, and adjusting the entire project review process based on the compliance assessment results, the process also includes: S50: Obtain the review progress of each application stage of the project, conduct a full-process review of the data upload time and corresponding uploaded review text for each stage of the review process, and obtain the review results of the stage process.

[0066] Specifically, in this embodiment, an RPA robot is used to monitor the review process of each application stage, automatically capture the data upload time parameters and uploaded review text of each stage of the review process, and obtain the stage process review results by identifying whether the review text of the corresponding stage is uploaded on time and whether the uploaded review text meets the requirements.

[0067] S60: Based on the review results of the phase process, determine whether the project deliverables data for the corresponding application phase meet the preset delivery indicators, extract the project deliverables data that do not meet the preset delivery indicators, and generate the corresponding phase review report.

[0068] Specifically, based on the review results of each stage, the project deliverables for the corresponding application stage are compared with the preset delivery indicators. The comparison results determine whether the project meets the preset delivery indicators, such as whether the paper has been accepted or the patent has been granted. Project deliverables that do not meet the expected delivery indicators are extracted, such as the indicators for unaccepted papers and ungranted patents, thus obtaining the corresponding stage review report.

[0069] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0070] In one embodiment, an intelligent project review system based on multi-agent fusion is provided, which corresponds one-to-one with the intelligent project review method based on multi-agent fusion described in the above embodiments. For example... Figure 8 As shown, this intelligent project review system based on multi-agent fusion includes an intelligent text review module, an intelligent problem analysis module, an intelligent report writing module, and a process automation module. Detailed descriptions of each functional module are as follows: The intelligent text review module is used to acquire project application texts, perform pre-review processing on the project application texts, and analyze the logical relationships of the content of the pre-reviewed project texts to obtain project text review data.

[0071] The intelligent problem analysis module is used to extract text features from project text review data and analyze potential problems in project application documents based on text features and corresponding content logic relationships to obtain project problem analysis data.

[0072] The intelligent report writing module is used to process project text review data and project problem analysis data from multiple dimensions according to a preset project review model, and to obtain a project review analysis report that conforms to the corresponding project review model.

[0073] The process automation module is used to perform key data verification processing on the project review and analysis report, assess the compliance of the entire project application based on the key data verification results, and adjust the entire project review process based on the compliance assessment results.

[0074] Preferably, the intelligent text review module specifically includes: The pre-review submodule is used to obtain project application texts, pre-review the project application texts according to preset text review conditions, and obtain pre-reviewed text data that meets the preset text review conditions.

[0075] The text association submodule is used to associate the pre-reviewed text data with each sub-project of the project, analyze the logical relationships between the content of each sub-project, and obtain logical analysis data between each sub-project.

[0076] The consistency analysis submodule is used to obtain the phased application data for each application stage of the project, perform consistency analysis of the application materials based on the phased application data and logical analysis data, and evaluate the trend of differences in application materials at different application stages.

[0077] The text annotation submodule is used to annotate project texts with differences in application materials based on the trend of differences in application materials, thereby obtaining project text review data for project applications.

[0078] Preferably, the intelligent problem analysis module specifically includes: The application prediction submodule is used to extract key text features from the project text review data and predict the project application results based on the text features and the corresponding content logic relationships.

[0079] The accuracy analysis submodule is used to obtain the actual application results of the corresponding projects, compare the project application prediction results with the actual application results according to the content logic, and analyze the accuracy of the project application prediction.

[0080] The problem identification submodule is used to identify potential problems in project application documents based on the accuracy of project application prediction, and to summarize them according to the corresponding project application stage to obtain project problem analysis data.

[0081] Preferably, before the intelligent report writing module processes the project text review data and project problem analysis data from multiple dimensions according to the preset project review model to obtain a project review analysis report that meets the corresponding report writing requirements, it also includes: The sample annotation submodule is used to obtain application case data of similar projects, use the application case data of similar projects as data samples for data training, and annotate the key text of the current project type to obtain text annotation data.

[0082] The model building submodule is used to analyze the text review conditions and model boundaries of the current project type based on the text annotation data, and to build a project review model.

[0083] Preferably, the intelligent report writing module specifically includes: The impact analysis submodule is used to perform impact analysis on project application based on the preset project review model, project text review data and project problem analysis data, and generate quality impact factors of project problems on the quality of project application.

[0084] The results prediction submodule is used to extract the corresponding key text data according to the preset report review conditions, predict the project review results of the key text data based on the quality influencing factors, and obtain the project review prediction results corresponding to the current report review conditions.

[0085] The material writing submodule is used to process key text data and corresponding project review prediction results according to preset report writing requirements, and to obtain a project review analysis report that meets the corresponding report writing requirements.

[0086] Preferably, the process automation module specifically includes: The data verification submodule is used to extract key data from the project review analysis report based on preset project review keywords, compare the key data with the corresponding project application standard parameters, and obtain the key data verification results.

[0087] The compliance assessment submodule is used to analyze the importance of each key data point in the entire project application based on the key data verification results, assess the compliance of the entire project application based on the importance analysis results, and generate a compliance assessment result.

[0088] The feedback adjustment submodule is used to feed back the compliance assessment results to each project application stage, adjust the text review conditions for each project application stage, and perform adaptive optimization for each project application stage.

[0089] Preferably, after the process automation module, it also includes: The process tracking module is used to obtain the stage review progress of each application stage of the project, and to conduct a full-process review of the data upload time and corresponding uploaded review text for each stage review process to obtain the stage process review results.

[0090] The stage review module is used to determine whether the project deliverables for the corresponding application stage meet the preset delivery indicators based on the review results of the stage progress, and to extract the project deliverables that do not meet the preset delivery indicators and generate the corresponding stage review report.

[0091] Specific limitations regarding the intelligent project review system based on multi-agent fusion can be found in the limitations of the intelligent project review method based on multi-agent fusion described above, and will not be repeated here. Each module in the aforementioned intelligent project review system based on multi-agent fusion can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0092] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 9 As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores intelligent project review data based on multi-agent fusion. The network interface communicates with external terminals via a network connection. When executed by the processor, the computer program implements an intelligent project review method based on multi-agent fusion.

[0093] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of an intelligent project review method based on multi-agent fusion.

[0094] Those skilled in the art will recognize that the units of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of the invention.

[0095] In the embodiments provided by the present invention, it should be understood that the division of units is only a logical functional division. In actual implementation, there may be other division methods, such as multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored.

[0096] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0097] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0098] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. An intelligent project auditing method based on multi-agent fusion, characterized in that, The method comprises: acquiring a project declaration text, pre-examining the project declaration text, and analyzing the content logical relationship of the pre-examined project text to obtain project text examination data; extracting the text features of the project text examination data, and analyzing potential problems in the project declaration document according to the text features and the corresponding content logical relationship to obtain project problem analysis data; according to a preset project audit model, performing multidimensional writing processing on the project text examination data and the project problem analysis data to obtain a project examination analysis report conforming to the corresponding project audit model; performing key data verification processing on the project examination analysis report, evaluating the overall project declaration compliance according to the key data verification result, and feeding back and adjusting the project examination whole process according to the compliance evaluation result. 2.The multi-agent fusion-based intelligent project auditing method according to claim 1, characterized in that, The acquisition of the project declaration text, the pre-examination of the project declaration text, and the analysis of the content logical relationship of the pre-examined project text to obtain project text examination data specifically comprises: acquiring a project declaration text, pre-examining the project declaration text according to a preset text examination condition, and obtaining text pre-examination data conforming to the preset text examination condition; associating the text pre-examination data with each sub-project of the project to analyze the content logical relationship between each sub-project and obtain logical analysis data between each sub-project; acquiring the staged declaration data of each declaration stage of the project, performing declaration material consistency analysis according to the staged declaration data and the logical analysis data, and evaluating the declaration material difference trend of different declaration stages; according to the declaration material difference trend, marking the project text with declaration material differences to obtain project text examination data of the project declaration. 3.The multi-agent fusion-based intelligent project auditing method according to claim 1, characterized in that, The extraction of the text features of the project text examination data and the analysis of potential problems in the project declaration document according to the text features and the corresponding content logical relationship to obtain project problem analysis data specifically comprises: performing key text feature extraction processing on the project text examination data, and predicting the project declaration result according to the text features and the corresponding content logical relationship to obtain a project declaration prediction result; acquiring the actual declaration result of the corresponding project, comparing the project declaration prediction result with the actual declaration result according to the content logical relationship, and analyzing the project declaration prediction accuracy; according to the project declaration prediction accuracy, identifying potential problems in the project declaration document, and summarizing according to the corresponding project declaration stage to obtain project problem analysis data. 4.The multi-agent fusion-based intelligent project auditing method according to claim 1, characterized in that, Before the multidimensional writing processing of the project text examination data and the project problem analysis data according to the preset project audit model to obtain the project examination analysis report conforming to the corresponding report writing requirement, it further comprises: acquiring the same type of project declaration case data, performing data training on the same type of project declaration case data as data samples, marking the key text of the current project type to obtain text marking data; According to the text labeling data, the text review conditions and model boundaries of the current project type are analyzed, and a project review model is constructed. 5.The multi-agent fusion-based intelligent project auditing method according to claim 4, characterized in that, According to the preset project review model, the project text review data and the project problem analysis data are subjected to multi-dimensional writing processing, and a project review analysis report meeting the corresponding report writing requirements is obtained, specifically including: According to the preset project review model, the project text review data and the project problem analysis data are subjected to project declaration influence analysis, and the quality influence factors of project problems on project declaration quality are generated; According to the preset report review conditions, the corresponding key text data is extracted, the key text data is subjected to project review result prediction according to the quality influence factors, and the project review prediction result corresponding to the current report review conditions is obtained; According to the preset report writing requirements, the key text data and the corresponding project review prediction result are subjected to material writing processing, and a project review analysis report meeting the corresponding report writing requirements is obtained. 6.The multi-agent fusion-based intelligent project auditing method according to claim 1, characterized in that, The key data verification processing of the project review analysis report is performed, the compliance of the entire project declaration is evaluated according to the key data verification result, and the project review whole process is adjusted according to the compliance evaluation result, specifically including: According to the preset project review keywords, the key data in the project review analysis report is extracted, the key data is compared with the corresponding project declaration standard parameters, and the key data verification result is obtained; According to the key data verification result, the importance of each key data in the entire project declaration is analyzed, the compliance of the entire project declaration is evaluated according to the importance analysis result, and the compliance evaluation result is generated; The compliance evaluation result is fed back to each project declaration stage, the text review conditions of each project declaration stage are adjusted, and each project declaration stage is subjected to adaptive optimization processing. 7.The multi-agent fusion-based intelligent project auditing method according to claim 1, characterized in that, After the key data verification processing of the project review analysis report, the compliance of the entire project declaration is evaluated according to the key data verification result, and the project review whole process is adjusted according to the compliance evaluation result, further including: Obtain the stage review process of each declaration stage of the project, review the material uploading time and the corresponding uploading review text of each stage review process, and obtain the stage process review result; According to the stage process review result, it is judged whether the project achievement delivery data of the corresponding declaration stage meets the preset delivery index, and the project achievement delivery data not meeting the preset delivery index is extracted, and the corresponding stage review report is generated.

8. An intelligent project auditing system based on multi-agent fusion, characterized in that, The system is applied to the intelligent project review method based on multi-agent fusion in any one of the above claims 1-7, and the system comprises: An intelligent text review module is used to obtain project declaration text, pre-review the project declaration text, and analyze the content logical relationship of the pre-reviewed project text to obtain project text review data; An intelligent problem analysis module is used to extract text features of the project text review data, and analyze potential problems in the project declaration document according to the text features and the corresponding content logical relationship to obtain project problem analysis data; The intelligent report writing module is configured to perform multidimensional writing processing on the project text review data and the project problem analysis data according to a preset project audit model, so as to obtain a project review analysis report conforming to the corresponding project audit model. The process automation module is configured to perform key data verification processing on the project review analysis report, evaluate the compliance of the entire project declaration according to a key data verification result, and feed back and adjust the entire project review process according to a compliance evaluation result.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the steps of the intelligent project audit method based on multi-agent fusion according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, the computer program comprising instructions that, when executed by a computer, cause the computer to perform the method of any one of claims 1 to 9. The computer program is executed by the processor to implement the steps of the intelligent project audit method based on multi-agent fusion according to any one of claims 1 to 7.

Images