Multi-model result processing method and system applied to document auditing

Abstract

The invention relates to a multi-model result processing method applied to official document auditing. The method comprises the following steps that S1, a user uploads an official document; s2, transferring the information in the step S1 to a task center for processing, creating an auditing task, splitting an official document into n fragments, and performing selective processing by using a plurality of models; s3, processing specific error types by the models according to the dominant fields of the models, summarizing processing results of the models, and uniformly converting the processing results into standard format output; s4, the front-end interface executes highlight display, error replacement and statistical operation according to the returned content, and finally a processing result is presented to the client; s5, recording each auditing result by the system; a multi-model result processing system applied to document auditing comprises a comprehensive application layer, a configuration management layer, a model processing layer and an infrastructure layer. The method has the advantage of realizing efficient and accurate auditing of the official document through various technologies of natural language processing and machine learning.

Description

Technical Field

[0001] The present invention relates to the technical field of information processing technology, and in particular to a method and system for processing multi-model results applied to official document review. Background Art

[0002] The traditional document review process often relies on manual review, which is not only time-consuming and labor-intensive, but also easily affected by personal experience and knowledge level, leading to inconsistencies and potential errors in the review results.

[0003] With the development of artificial intelligence technology, especially the advancement of natural language processing (NLP) technology, it has become possible to use models to automate document review. However, a single model is often difficult to cover all types of errors in document review. Therefore, how to effectively integrate the results of multiple models to comprehensively and accurately review documents has become a problem that needs to be solved urgently. Summary of the invention

[0004] In view of the above-mentioned deficiencies in the prior art, the technical problem to be solved by this patent application is how to provide a method and system for processing multi-model results applied to official document review, which realizes efficient and accurate review of official documents through a variety of technologies such as natural language processing and machine learning.

[0005] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0006] A method for processing multi-model results applied to official document review includes the following steps: S1: a user uploads an official document; S2: the information flow in step S1 is transferred to a task center for processing, an audit task is created, the official document is split into n fragments, and multiple models are used to select and process; S3: each model processes a specific error type according to its advantage field, summarizes the processing results of each model, and uniformly converts them into a standard format for output; S4: a front-end interface performs highlighting, error replacement and statistical operations based on the returned content, and finally presents the processing results to the customer; S5: the system records the results of each audit and allows the user to review the past audit history and modification records at any time.

[0007] As an optimization, in step S1, when the user uploads the document to the system, he / she needs to check the model capabilities he / she wants to use to create personalized user review requirements.

[0008] As an optimization, in step S2, the model selected when processing the official document includes but is not limited to a semantic analysis model, a grammar checking model, and a standard terminology detection model, and the user can arbitrarily select at least one of the models for use.

[0009] As an optimization, in step S3, the processing results of each model are summarized to generate an audit report, which includes information on the identified error type, location, error description, and recommended modification plan.

[0010] In this embodiment, in step S4, the audit report is displayed to the user to facilitate the user's review and operation. The user can modify the official document according to the suggestions in the audit report to improve the quality and standardization of the official document.

[0011] As an optimization, step S5 also has the function of statistically analyzing the types of user review errors. By recording and analyzing the types and numbers of errors encountered by users during the review process, a detailed statistical report can be generated to help users understand the difficulties and error-prone points of official document review. The statistical report includes key information such as error type distribution, error frequency, and error causes, providing a reference for users' subsequent improvements and optimizations.

[0012] A processing system for multi-model results applied to official document review comprises a comprehensive application layer, a configuration management layer, a model processing layer and an infrastructure layer; the comprehensive application layer is used for displaying a page from a user submitting a review requirement to selecting model capabilities included in the review requirement; the configuration management layer is used for configuring different models for processing various error types during the review process, and setting parameters of each model to optimize model processing; the model processing layer uniformly processes responses to various model results, summarizes errors that can be identified by multiple models, and can best integrate multiple model capabilities; the infrastructure layer is used to provide operating platform support for the system.

[0013] As an optimization, the configuration management layer includes a user interaction module, a data model configuration management module, a user document operation module, an official document detection module, an official document writing module, and a data analysis module.

[0014] As an optimization, the data center layer includes a data source management module, a data model configuration management module, a collection requirement management module, a collection template management module, a collection task release management module, a collection task reporting management module, a collection task review management module, a collection task monitoring module and a circulation mode management module; the data center layer includes a data source management module for managing the source and format of official document data; the data model configuration management module is used to configure and optimize the data model required for model processing; the collection requirement management module is used to define and manage the user's review requirements; the collection template management module is used to create and manage templates for official document review; the collection task release management module is used to release and manage official document review tasks; the collection task reporting management module is used to manage and monitor the reporting progress of tasks; the collection task review management module is used to review and manage reported tasks; the collection task monitoring module is used to monitor the progress and status of official document review tasks in real time; the circulation mode management module is used to define and manage the circulation mode and process of official document review.

[0015] As an optimization, the model processing layer selects a suitable model to review the official document according to the configuration of the configuration management layer and the user's review requirements, and returns the review results to the comprehensive application layer for display and processing; the infrastructure layer ensures that the official document review system can run stably and securely, and provides efficient data processing and storage capabilities.

[0016] The present invention has the following advantages:

[0017] (1) By integrating the results of multiple models, the present invention can comprehensively and accurately review official documents, thereby improving the accuracy and efficiency of the review;

[0018] (2) The system supports official documents in various formats and can convert the processing results into a specific standard format for output, making it convenient for users to view and operate;

[0019] (3) The front-end interface is user-friendly and can intuitively display the error locations and modification suggestions in the document, thus improving the user experience;

[0020] (4) The system records the results and modification records of each review, providing users with convenient review and backtracking functions. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 The present invention is a flowchart of a method for processing multi-model results applied to official document review.

[0022] Figure 2 This is a framework diagram of a multi-model result processing system for official document review described in the present invention. DETAILED DESCRIPTION

[0023] The present invention is further described in detail below in conjunction with the accompanying drawings. In the description of the present invention, it should be understood that the directions or positional relationships indicated by directional words such as "upper, lower" and "top, bottom" are usually based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description. Unless otherwise stated, these directional words do not indicate or imply that the devices or components referred to must have a specific direction or be constructed and operated in a specific direction, and therefore cannot be understood as limiting the scope of protection of the present invention; the directional words "inside and outside" refer to the inside and outside relative to the contours of each component itself.

[0024] Example 1

[0025] like Figure 1 As shown, a method for processing multi-model results applied to official document review includes the following steps: S1: the user uploads the official document; S2: the information flow in step S1 is transferred to the task center for processing, an audit task is created, the official document is split into n fragments, and multiple models are used to select and process; S3: each model processes a specific error type according to its advantage field, summarizes the processing results of each model, and uniformly converts them into a standard format output; S4: the front-end interface performs highlighting, error replacement and statistical operations based on the returned content, and finally presents the processing results to the customer; S5: the system records the results of each audit and allows users to review past audit history and modification records at any time.

[0026] In this embodiment, in step S1, when the user uploads the document to the system, the user needs to check the model capabilities that he wants to use to create personalized user review requirements.

[0027] Specifically, when the user uploads the official document, the methods that can be used include but are not limited to document import, copy and paste.

[0028] Specifically, the uploaded files support multiple formats, including but not limited to Word and PDF.

[0029] Specifically, after uploading the official document, the document format is verified to ensure the integrity of the document content and the standardization of the format. Subsequently, the system pre-processes the document, including removing irrelevant information, text cleaning (as well as the basic work of word segmentation and part-of-speech tagging, to prepare for subsequent AI model processing.

[0030] Specifically, when configuring and selecting model capabilities, including but not limited to semantic analysis models, grammar checking models, and standard terminology detection models, the document review process and the models used are dynamically adjusted according to the user's configuration to meet the review needs of different users.

[0031] In this embodiment, in step S2, the model selected when processing the official document includes but is not limited to a semantic analysis model, a grammar checking model, and a standard terminology detection model, and the user can arbitrarily select at least one of the models for use.

[0032] Specifically, the semantic analysis model is responsible for analyzing the semantic content of official documents, identifying key information such as themes, viewpoints, and logical relationships in official documents, and judging whether there are semantic errors or inappropriateness in official documents by comparing the content of official documents with a preset semantic library or standard; the grammar checking model is responsible for checking the grammatical correctness of official documents, including sentence structure, tense, voice, and subject-verb agreement. By comparing the content of official documents with a grammar rule library, grammatical errors are identified and modification suggestions are made; the standard terminology detection model: This model is responsible for detecting the use of standard terminology in official documents, ensuring that the terminology in official documents is used accurately and in a standardized manner, and identifying inappropriate use of terminology and making modification suggestions by comparing the content of official documents with a standard terminology library.

[0033] Specifically, it also includes format specification checking model and logical consistency checking model to conduct a more comprehensive review of official documents.

[0034] In this embodiment, in step S3, the processing results of each model are summarized to generate an audit report, which includes information on the identified error type, location, error description, and recommended modification plan.

[0035] Specifically, the standard format of the output is compatible with a variety of content forms, including but not limited to errors identified in tables and images.

[0036] In this embodiment, in step S4, the audit report is displayed to the user for the convenience of the user to review and operate. The user can modify the official document according to the suggestions in the audit report to improve the quality and standardization of the official document.

[0037] In this embodiment, step S5 also has the function of statistically analyzing the types of user review errors. By recording and analyzing the types and numbers of errors encountered by users during the review process, a detailed statistical report can be generated to help users understand the difficulties and error-prone points of official document review. The statistical report includes key information such as error type distribution, error frequency, and error causes, providing a reference for users' subsequent improvements and optimizations.

[0038] Example 2

[0039] like Figure 2 As shown, a multi-model result processing system applied to official document review includes a comprehensive application layer, a configuration management layer, a model processing layer and an infrastructure layer;

[0040] The comprehensive application layer is used for users to submit audit requirements and select the model capabilities included in the audit requirements. The configuration management layer is used to configure different models for processing various error types during the audit process, and set the parameters of each model to optimize the model processing; the model processing layer unifies the responses of various model results, summarizes errors that can be identified by multiple models, and best combines multiple model capabilities; the infrastructure layer is used to provide operating platform support for the system.

[0041] In this embodiment, the configuration management layer includes a user interaction module, a data model configuration management module, a user document operation module, a document detection module, a document writing module, and a data analysis module.

[0042] In this embodiment, the data center layer includes a data source management module, a data model configuration management module, a collection demand management module, a collection template management module, a collection task release management module, a collection task reporting management module, a collection task review management module, a collection task monitoring module and a circulation mode management module;

[0043] The data center layer includes a data source management module for managing the source and format of official document data; the data model configuration management module is used to configure and optimize the data model required for model processing; the collection requirement management module is used to define and manage the user's review requirements; the collection template management module is used to create and manage templates for official document review; the collection task release management module is used to release and manage official document review tasks; the collection task reporting management module is used to manage and monitor the reporting progress of tasks; the collection task review management module is used to review and manage reported tasks; the collection task monitoring module is used to monitor the progress and status of official document review tasks in real time; the circulation mode management module is used to define and manage the circulation mode and process of official document review.

[0044] In this embodiment, the model processing layer selects a suitable model to review the official document according to the configuration of the configuration management layer and the user's review requirements, and returns the review results to the comprehensive application layer for display and processing; the infrastructure layer ensures that the official document review system can run stably and securely, and provides efficient data processing and storage capabilities.

[0045] Specifically, the system calls the corresponding AI model in the model storage layer to review documents according to the preset review rules or user-defined review requirements. In order to optimize the model review rate and reduce user review time, the system has taken the following measures: Parallel processing: The system supports multiple models to process document fragments in parallel to improve processing efficiency. By reasonably allocating computing resources, the system can run multiple models at the same time to achieve parallel processing of document review; Intelligent scheduling: The system intelligently schedules document fragments to different models for processing based on the load and processing speed of the model. By dynamically adjusting the model allocation strategy, the system can optimize the utilization of the model and improve the overall review rate; Cache mechanism: The system uses a cache mechanism to cache commonly used document fragments and model processing results to reduce repeated calculations and improve processing speed.

[0046] Specifically, the system also has the function of statistically analyzing the error types of the user review interface. By recording and analyzing the types and numbers of errors encountered by users during the review process, the system can generate detailed statistical reports to help users understand the difficulties and error-prone points of official document review. The statistical reports include key information such as error type distribution, error frequency, and error causes, providing a reference for users' subsequent improvements and optimizations.

[0047] Specifically, a user feedback mechanism has been introduced. Users can provide manual feedback on errors not recognized by the system on the audit report page. In the original audit text, users can mark the error areas not recognized by the system, select the error type, and click the Submit button to send the feedback information to the system backend. After receiving the user feedback, the system backend will automatically learn and update the model to improve the ability to recognize similar errors in the future.

[0048] Specifically, in order to solve the problem of false alarms, a black and white list mechanism is introduced. Users can control the system's recognition and processing of certain specific content by setting black and white lists.

[0049] Finally, it should be noted that those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for processing multi-model results applied to official document review, characterized in that: The following steps are involved: S1: The user uploads the document; S2: The information flow in step S1 is transferred to the task center for processing, where a review task is created, the document is split into n pieces, and multiple models are used to select and process them; S3: Each model processes a specific error type according to its advantage area, summarizes the processing results of each model, and converts them into a standard format for output; S4: The front-end interface performs highlighting, error replacement and statistical operations based on the returned content, and finally presents the processing results to the customer; S5: The system records the results of each audit and allows users to review past audit history and modification records at any time.

2. According to claim 1, a method for processing multi-model results applied to official document review is characterized in that: In step S1, when the user uploads the document to the system, he / she needs to check the model capabilities he / she wants to use to create personalized user review requirements.

3. The method for processing multi-model results applied to document review according to claim 2 is characterized in that: In step S2, the model selected when processing the document includes but is not limited to a semantic analysis model, a grammar checking model, and a standard terminology detection model, and the user can arbitrarily select at least one of the models for use.

4. The method for processing multi-model results applied to document review according to claim 3 is characterized in that: In step S3, the processing results of each model are summarized to generate an audit report, which includes information on the identified error type, location, error description, and recommended modification plan.

5. The method for processing multi-model results applied to official document review according to claim 4 is characterized in that: In step S4, the audit report is displayed to the user for easy review and operation by the user. The user can modify the official document according to the suggestions in the audit report to improve the quality and standardization of the official document.

6. The method for processing multi-model results applied to official document review according to claim 5 is characterized in that: In step S5, there is also the function of statistically analyzing the types of user review errors. By recording and analyzing the types and numbers of errors encountered by users during the review process, a detailed statistical report can be generated to help users understand the difficulties and error-prone points of official document review. The statistical report includes key information such as error type distribution, error frequency, and error causes, providing a reference for users' subsequent improvements and optimizations.

7. A multi-model result processing system applied to official document review, characterized in that: It includes comprehensive application layer, configuration management layer, model processing layer and infrastructure layer; The comprehensive application layer is used for users to submit audit requirements and display the page for selecting model capabilities included in the audit requirements; The configuration management layer is used to configure different models for processing various error types during the audit process, and set parameters of each model to optimize model processing; The model processing layer uniformly processes the responses of various model results, summarizes the errors that can be identified by multiple models, and is the best at integrating the capabilities of multiple models; The infrastructure layer is used to provide operating platform support for the system.

8. A multi-model result processing system applied to official document review, characterized in that: The configuration management layer includes a user interaction module, a data model configuration management module, a user document operation module, an official document detection module, an official document writing module, and a data analysis module.

9. A multi-model result processing system applied to official document review, characterized in that: The data center layer includes a data source management module, a data model configuration management module, a collection demand management module, a collection template management module, a collection task release management module, a collection task reporting management module, a collection task review management module, a collection task monitoring module and a circulation mode management module; The data center layer includes a data source management module for managing the source and format of official document data; the data model configuration management module is used to configure and optimize the data model required for model processing; the collection requirement management module is used to define and manage the user's review requirements; the collection template management module is used to create and manage templates for official document review; the collection task release management module is used to release and manage official document review tasks; the collection task reporting management module is used to manage and monitor the reporting progress of tasks; the collection task review management module is used to review and manage reported tasks; the collection task monitoring module is used to monitor the progress and status of official document review tasks in real time; the circulation mode management module is used to define and manage the circulation mode and process of official document review.

10. A multi-model result processing system applied to official document review, characterized in that: The model processing layer selects a suitable model to review the official document according to the configuration of the configuration management layer and the user's review requirements, and returns the review results to the comprehensive application layer for display and processing; the infrastructure layer ensures that the official document review system can run stably and securely, and has efficient data processing and storage capabilities.

Images