Multi-level multi-model collaborative compliance detection system and method based on large model

Abstract

The invention provides a multilevel multi-model collaborative compliance detection system and method based on a large model, and the system comprises a content analysis and verification module which is used for extracting text information in to-be-detected content and correcting errors in the extracted content; the compliance detection module is used for detecting compliance; the consistency detection module is used for detecting the consistency; and the detection result rechecking module is used for rechecking and checking the detection result based on the compliance criterion. According to the technical scheme, compliance detection can be carried out on various types of materials, and compliance detection capability is flexibly constructed for compliance detection of specific professions and scene vertical fields; the information consistency of specific professional and scene contents can be flexibly detected.

Description

Background Art

[0002] Compliance testing and review of materials is an important step for most organizations and companies before releasing official materials. Using traditional rules or deep learning models to assist manual testing is a common technical solution. However, it also has problems such as a single form of detectable content, poor understanding of long texts and context, inflexible consistency testing of the content information being tested, and weak semantic generalization capabilities.

[0003] Prior art (invention patent with announcement number CN118228250A) The present invention discloses a data compliance detection method and system based on a large language model and AI Agent, which relates to the field of data detection technology. The present invention can realize the automation and efficiency of data compliance detection, and improve the accuracy and adaptability of data compliance detection. However, in actual situations, first, the materials and requirements for compliance detection required by users often come from various types of material sources, such as text, voice, video, pictures, etc., and not just data compliance; second, it is often necessary to have compliance detection capabilities in specific professional and scenario vertical fields, which is not easy to meet with this solution; third, it is often necessary to have the ability to detect the consistency of content information, which this solution does not have. Summary of the invention

[0004] The present application provides a multi-level and multi-model collaborative compliance detection system and method based on a large model to improve audit efficiency and unify the compliance audit level.

[0005] In the first aspect, a multi-level and multi-model collaborative compliance detection system based on a large model is provided, comprising:

[0006] Content parsing and verification module, used to extract text information from the content to be checked and correct errors in the extracted content;

[0007] A compliance detection module, used for detecting compliance;

[0008] The consistency detection module is used to detect the consistency.

[0009] In the above technical solution, a content parsing and verification module is set up to extract text information from the content to be checked and correct errors in the extracted content; a compliance detection module is used to detect compliance; a consistency detection module is used to detect consistency; it is capable of performing compliance detection on various types of materials, and flexibly building compliance detection capabilities for compliance detection in specific professional and scenario vertical fields; it is capable of flexibly detecting information consistency in specific professional and scenario content.

[0010] In a specific embodiment, it also includes:

[0011] The detection rule formulation and compliance model training module is used to build detection rules and train deep learning models for specific detection tasks and fine-tune low-parameter vertical field large models;

[0012] The rules and terms knowledge base module is used to build a rules and terms knowledge base.

[0013] In a specific embodiment, it also includes:

[0014] The test result review module is used to review and supplement the test results based on compliance criteria;

[0015] The audit data management and processing module is used for audit data management and traceability as well as marking and processing audit data.

[0016] In a specific embodiment, it also includes:

[0017] The compliance test report generation module is used to generate compliance test reports.

[0018] In a specific embodiment, it also includes:

[0019] The manual review and error correction module is used by professionals to review, correct and mark the generated compliance test reports.

[0020] In the second aspect, a multi-level and multi-model collaborative compliance detection method based on a large model is provided, comprising the following steps:

[0021] Utilize the content analysis and verification module to extract text information from the content to be checked and correct errors in the extracted content;

[0022] Use the compliance detection module to detect compliance;

[0023] The consistency detection module is used to detect the consistency.

[0024] In the above technical solution, a content parsing and verification module is set up to extract text information from the content to be checked and correct errors in the extracted content; a compliance detection module is used to detect compliance; a consistency detection module is used to detect consistency; it is capable of performing compliance detection on various types of materials, and flexibly building compliance detection capabilities for compliance detection in specific professional and scenario vertical fields; it is capable of flexibly detecting information consistency in specific professional and scenario content.

[0025] In a specific embodiment, it also includes:

[0026] Use the detection rule formulation and compliance model training module to build detection rules and train deep learning models for specific detection tasks and fine-tune low-parameter vertical field large models;

[0027] Use the rules and terms knowledge base module to build a rules and terms knowledge base.

[0028] In a specific embodiment, it also includes:

[0029] Use the test result review module to review and supplement the test results based on compliance criteria;

[0030] Use the audit data management and processing module to manage and record audit data as well as mark and process audit data.

[0031] In a specific embodiment, it also includes:

[0032] Generate compliance test reports using the compliance test report generation module.

[0033] In a specific embodiment, it also includes:

[0034] Professionals use the manual review and error correction module to review, correct and mark the generated compliance test reports. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A structural block diagram of a multi-level and multi-model collaborative compliance detection system based on a large model provided in an embodiment of the present application;

[0036] Figure 2 A flowchart of a multi-level and multi-model collaborative compliance detection method based on a large model provided in an embodiment of the present application;

[0037] Figure 3 A specific flow chart of the multi-level and multi-model collaborative compliance detection method based on a large model provided in an embodiment of the present application. DETAILED DESCRIPTION

[0038] The present application is further described in detail below through the accompanying drawings and embodiments. Through these descriptions, the characteristics and advantages of the present application will become clearer and more specific.

[0039] The word "exemplary" is used exclusively herein to mean "serving as an example, embodiment, or illustration." Any embodiment described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments. Although various aspects of the embodiments are shown in the drawings, the drawings are not necessarily drawn to scale unless otherwise noted.

[0040] In addition, the technical features involved in different embodiments of the present application described below can be combined with each other as long as they do not conflict with each other.

[0041] To facilitate understanding of the multi-level multi-model collaborative compliance detection system and method based on a large model provided in the embodiment of the present application, its application scenario is first explained. The multi-level multi-model collaborative compliance detection system and method based on a large model provided in the embodiment of the present application are used to improve audit efficiency and unify the level of compliance audit. Compliance detection and review of materials are important steps before most institutions and enterprises release formal materials. The use of traditional rules or deep learning models to assist manual detection is a relatively common technical solution. However, it also has problems such as a single form of detectable content, poor understanding of long texts and contextual contexts, inflexible consistency detection of the detected content information, and weak semantic generalization ability. Prior art (invention patent with announcement number CN118228250A) The present invention discloses a data compliance detection method and system based on a large language model and an AI Agent, which relates to the field of data detection technology. The present invention can realize the automation and efficiency of data compliance detection, and improve the accuracy and adaptability of data compliance detection. However, in actual situations, first, the materials and requirements for compliance testing required by users often come from various types of sources, such as text, voice, video, pictures, etc., and not just data compliance; second, it is often necessary to have compliance testing capabilities in specific professional and scenario vertical fields, which is not easy to meet with this solution; third, it is often necessary to have the ability to detect the consistency of content information, which this solution does not have. For this reason, the embodiment of the present application provides a multi-level and multi-model collaborative compliance testing system and method based on a large model to improve audit efficiency and unify the level of compliance audits. The following is a detailed description of this with specific drawings and examples.

[0042] refer to Figures 1 to 3 , Figure 1 A structural block diagram of a multi-level and multi-model collaborative compliance detection system based on a large model provided in an embodiment of the present application; Figure 2 A flowchart of a multi-level and multi-model collaborative compliance detection method based on a large model provided in an embodiment of the present application; Figure 3 A specific flow chart of the multi-level and multi-model collaborative compliance detection method based on a large model provided in an embodiment of the present application.

[0043] exist Figure 1 In the embodiment of the present application, a multi-level multi-model collaborative compliance detection system based on a large model is provided, including:

[0044] Content parsing and verification module, used to extract text information from the content to be checked and correct errors in the extracted content;

[0045] A compliance detection module, used for detecting compliance;

[0046] The consistency detection module is used to detect the consistency.

[0047] In the above technical solution, a content parsing and verification module is set up to extract text information from the content to be checked and correct errors in the extracted content; a compliance detection module is used to detect compliance; a consistency detection module is used to detect consistency; it is capable of performing compliance detection on various types of materials, and flexibly building compliance detection capabilities for compliance detection in specific professional and scenario vertical fields; it is capable of flexibly detecting information consistency in specific professional and scenario content.

[0048] In a specific embodiment, it also includes:

[0049] The detection rule formulation and compliance model training module is used to build detection rules and train deep learning models for specific detection tasks and fine-tune low-parameter vertical field large models;

[0050] The rules and terms knowledge base module is used to build a rules and terms knowledge base.

[0051] In a specific embodiment, it also includes:

[0052] The test result review module is used to review and supplement the test results based on compliance criteria. ;

[0053] The audit data management and processing module is used for audit data management and traceability as well as marking and processing audit data.

[0054] In a specific embodiment, it also includes:

[0055] The compliance test report generation module is used to generate compliance test reports.

[0056] In a specific embodiment, it also includes:

[0057] The manual review and error correction module is used by professionals to review, correct and mark the generated compliance test reports.

[0058] exist Figure 2 In the embodiment of the present application, a multi-level multi-model collaborative compliance detection method based on a large model is provided, comprising the following steps:

[0059] Utilize the content analysis and verification module to extract text information from the content to be checked and correct errors in the extracted content;

[0060] Use the compliance detection module to detect compliance;

[0061] The consistency detection module is used to detect the consistency.

[0062] In the above technical solution, a content parsing and verification module is set up to extract text information from the content to be checked and correct errors in the extracted content; a compliance detection module is used to detect compliance; a consistency detection module is used to detect consistency; it is capable of performing compliance detection on various types of materials, and flexibly building compliance detection capabilities for compliance detection in specific professional and scenario vertical fields; it is capable of flexibly detecting information consistency in specific professional and scenario content.

[0063] In a specific embodiment, it also includes:

[0064] Use the detection rule formulation and compliance model training module to build detection rules and train deep learning models for specific detection tasks and fine-tune low-parameter vertical field large models;

[0065] Use the rules and terms knowledge base module to build a rules and terms knowledge base.

[0066] In a specific embodiment, it also includes:

[0067] Use the test result review module to review and supplement the test results based on compliance criteria. ;

[0068] Use the audit data management and processing module to manage and record audit data as well as mark and process audit data.

[0069] In a specific embodiment, it also includes:

[0070] Generate compliance test reports using the compliance test report generation module.

[0071] In a specific embodiment, it also includes:

[0072] Professionals use the manual review and error correction module to review, correct and mark the generated compliance test reports.

[0073] Specifically, refer to Figure 3 , a multi-level and multi-model collaborative compliance detection method based on a large model, identifies the content to be detected through a multi-modal large model, and uses rules, deep learning models, vertical field large models, general large models and other technical detection capabilities to complete compliance detection in specific professional fields, helping professional auditors improve audit efficiency and unify compliance audit levels. Specifically, it includes the following steps:

[0074] Step 1: Use the "Detection Rule Development and Compliance Model Training Module"

[0075] 1) Based on compliance rules, past experience and data, build detection rules and train deep learning models for specific detection tasks; 2) Based on the professional field data accumulated in the past, fine-tune the low-parameter vertical field large model.

[0076] Step 2: Use the "rules and terms knowledge base module" to collect and build the required matching rules and terms knowledge base.

[0077] Step 3: Get the content to be tested, which can be text, voice, image, video, etc.

[0078] Step 4: Using the “Content Parsing and Verification Module”, 1) use the multimodal big model to extract text information from the content to be checked; 2) use the big model-based AI Agent to correct obvious errors in the extracted content.

[0079] Step 5: Use the "Compliance Detection Module" to detect compliance based on detection rules, deep learning models, fine-tuned vertical field low-parameter large models, etc.

[0080] Step 6: Use the "consistency detection module" to combine detection rules, rule clause knowledge base, and large model RAG technology to detect consistency.

[0081] Step 7: Use the "Test Result Review Module" to review previous test results and identify and fill in gaps based on compliance criteria using a high-parameter general large model.

[0082] Step 8: Use the "Compliance Test Report Generation Module" to generate a compliance test report based on previous test results using a large model for manual review and record keeping.

[0083] Step 9: Using the “manual review and error correction module”, professionals review, correct and mark the compliance detection report generated by the large model in order to officially release the final compliance audit report.

[0084] Step 10: Utilize the “Audit Data Management and Processing Module” to 1) manage and record audit data; 2) mark and process audit data for further iterative optimization of the model.

[0085] In this embodiment, it has the compliance detection capability of multimodal scenarios; it can build a multi-level and multi-model collaborative detection method to meet the compliance and consistency detection needs in specific professional and scenario vertical fields; and build a multi-level and multi-model collaborative compliance detection system based on a multimodal large model.

[0086] In the above technical scheme, in the above technical scheme, a content parsing and verification module is set up to extract text information from the content to be checked and correct errors in the extracted content; a compliance detection module is used to detect compliance; a consistency detection module is used to detect consistency; a test result review module is used to review and supplement the test results based on compliance criteria; it can perform compliance testing on various types of materials, and flexibly build compliance testing capabilities for compliance testing in specific professional and scenario vertical fields; it can flexibly detect the information consistency of specific professional and scenario content.

[0087] Those skilled in the art will appreciate that the present application may be implemented as a system, method or computer program product.

[0088] Therefore, the present disclosure may be specifically implemented in the following forms, namely: it may be completely hardware, it may be completely software (including firmware, resident software, microcode, etc.), or it may be a combination of hardware and software, generally referred to herein as a "circuit", "module" or "system". In addition, in some embodiments, the present application may also be implemented in the form of a computer program product in one or more computer-readable media, and the computer-readable medium may contain computer-readable program code.

[0089] Any combination of one or more computer-readable media can be used. Computer-readable media can be computer-readable signal media or computer-readable storage media. Computer-readable storage media can be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or devices, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: electrical connections with one or more wires, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In this document, computer-readable storage media can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it.

[0090] Although the embodiments of the present application have been shown and described above, it is understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art can change, modify, replace and modify the above embodiments within the scope of the present application. On this basis, a variety of replacements and improvements can be made to the present application, all of which fall within the scope of protection of the present application.

Claims

1. A multi-level and multi-model collaborative compliance detection system based on a large model, characterized in that: include: Content parsing and verification module, used to extract text information from the content to be checked and correct errors in the extracted content; A compliance detection module, used for detecting compliance; The consistency detection module is used to detect the consistency.

2. The multi-level and multi-model collaborative compliance detection system based on a large model according to claim 1 is characterized in that: Also includes: The detection rule formulation and compliance model training module is used to build detection rules and train deep learning models for specific detection tasks and fine-tune low-parameter vertical field large models; The rules and terms knowledge base module is used to build a rules and terms knowledge base.

3. The multi-level and multi-model collaborative compliance detection system based on a large model according to claim 2 is characterized in that: Also includes: The test result review module is used to review and supplement the test results based on compliance criteria. ； The audit data management and processing module is used for audit data management and traceability as well as marking and processing audit data.

4. The multi-level and multi-model collaborative compliance detection system based on a large model according to claim 3 is characterized in that: Also includes: The compliance test report generation module is used to generate compliance test reports.

5. The multi-level and multi-model collaborative compliance detection system based on a large model according to claim 4 is characterized in that: Also includes: The manual review and error correction module is used by professionals to review, correct and mark the generated compliance test reports.

6. A multi-level and multi-model collaborative compliance detection method based on a large model, characterized in that: The following steps are involved: Utilize the content analysis and verification module to extract text information from the content to be checked and correct errors in the extracted content; Use the compliance detection module to detect compliance; The consistency detection module is used to detect the consistency.

7. The multi-level and multi-model collaborative compliance detection method based on a large model according to claim 6 is characterized in that: Also includes: Use the detection rule formulation and compliance model training module to build detection rules and train deep learning models for specific detection tasks and fine-tune low-parameter vertical field large models; Use the rules and terms knowledge base module to build a rules and terms knowledge base.

8. The multi-level and multi-model collaborative compliance detection method based on a large model according to claim 7 is characterized in that: Also includes: Use the test result review module to review and supplement the test results based on compliance criteria; Use the audit data management and processing module to manage and record audit data as well as mark and process audit data.

9. The multi-level and multi-model collaborative compliance detection method based on a large model according to claim 8 is characterized in that: Also includes: Generate compliance test reports using the compliance test report generation module.

10. The multi-level and multi-model collaborative compliance detection method based on a large model according to claim 9 is characterized in that: Also includes: Professionals use the manual review and error correction module to review, correct and mark the generated compliance test reports.

Images