Recording implementation method based on multi-dimensional auditing agent execution process log

Abstract

The invention discloses a recording implementation method based on a multi-dimensional auditing agent execution process log, and the method specifically comprises the following steps: S1, auditing uploaded document data through an auditing system, and auditing all auditing points of a document; s2, after each auditing point of the document is audited, the auditing system outputs an auditing result to each auditing point, outputs a final auditing result according to all the auditing results of the document, and outputs text logs of all the auditing points of the document at the same time; s3, when the auditing of the auditing point fails, the auditing system stores the auditing field data of the auditing point into a database; and S4, after the auditing of the auditing point fails, performing manual secondary auditing confirmation, and outputting a final auditing result. According to the method, data recording and storage are carried out on the auditing process in multiple dimensions, and the full view of the auditing execution process is restored; meanwhile, reservation of original data on the site is audited, and rapid judgment and intervention on failed audit points are achieved manually; and responsibility investigation or audit checking of subsequent data is facilitated.

Description

Technical Field

[0001] This invention belongs to the field of data processing technology, specifically relating to a method for recording the execution process log of a multi-dimensional auditing agent. Background Technology

[0002] The intelligent text review system is a business system built upon cutting-edge artificial intelligence technology and professional review knowledge from various fields. It assists various business personnel in completing complex text review tasks. The system employs large language model technology and a unique rule engine and review engine technology to assist in a series of tasks including review, archiving, and analysis. It helps clients control risks, improve efficiency, reduce costs, and optimize the review process.

[0003] Traditional text review systems only record the final results of each review point under the document and the final review result of the entire document. The recording method is fast and efficient, but it is difficult to restore the whole picture of the review execution process. It is difficult to use for business compliance audit, investigation of anomalies when the review fails, or statistical analysis of the reasons for the review failure afterward. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the existing technology and provide a method for recording the execution process log of a multi-dimensional auditing agent. The method of this invention records and retains data of the auditing process from multiple dimensions, which improves the accuracy of log recording and clearly restores the whole picture of the auditing execution process. At the same time, the retention of the original data at the audit site enables the rapid judgment and intervention of human personnel at the failed audit points, which facilitates the subsequent data traceability or audit verification.

[0005] To achieve the above-mentioned objectives, the technical solution provided by this invention patent is as follows:

[0006] A method for recording the execution process log of a multi-dimensional auditing agent, the method specifically includes the following steps:

[0007] S1, the audit system audits the uploaded document data and audits all audit points of the document according to the business rules for document audit;

[0008] S2, after each audit point of the document is completed, the audit system outputs the audit result for each audit point. The audit system outputs the final audit result based on all the audit results of the document, and at the same time outputs the text log of all audit points of the document and stores the text log in the database;

[0009] S3, When an audit point fails, the audit system stores the audit site data of that audit point in the database;

[0010] S4. If the audit point fails, a second manual audit is conducted. The second audit is conducted based on the text log and audit site data corresponding to the audit point, and the final audit result is output.

[0011] Furthermore, the text log of each audit point includes the audit result, audit reason, audit flow, physical processing method, audit start time, audit end time, and audit timeliness; when each audit point is audited, the output text log also includes the detailed time of text log generation.

[0012] Furthermore, the audit results output by each audit point include audit passed and audit failed. Audit failed includes rejection, transfer to human agent, and communication. When all audit results output by the audit points are audit passed during document audit, the document's audit result is audit passed. When an audit failed result appears in the audit point output during document audit, the document's audit result is audit failed.

[0013] Furthermore, after the document's audit result is output as an audit failure, the specific audit failure result is output according to the priority of the failed audit point. The priority of audit failure is, in order, rejection, transfer to manual processing, and communication.

[0014] Furthermore, when the audit point fails, the audit site data of the audit point includes the original bill of lading information, attachment information, and ledger information that caused the audit failure.

[0015] Furthermore, the manual secondary review and confirmation specifically involves: when a document review fails, the manual reviewer determines the reason for the failure of the review point based on the review reason and the data at the review site; and based on the reason for the failure of the review point, the manual reviewer selects the corresponding reason for the failure of the review point and modifies the review result accordingly, while recording the dimensional information of the manual operation.

[0016] Furthermore, the reasons for the failure of the audit point include AI parsing errors, unreasonable rule configuration, missing bill of lading attachments, redundant bill of lading attachments, and incorrect bill of lading information; the dimension information of the manual operation includes the time of the audit point modification, operation type, operation content, and operator information.

[0017] Based on the above technical solution, the recording method of the execution process log of the intelligent agent based on multi-dimensional auditing patent has achieved the following technical advantages through practical application:

[0018] 1. The present invention provides a method for recording the execution process log of a multi-dimensional auditing agent. By recording and retaining data of the auditing process from multiple dimensions, the accuracy of log recording is improved, and the overall picture of the auditing execution process is clearly restored. At the same time, the retention of the original data at the audit site enables rapid human judgment and intervention at failed audit points, which facilitates subsequent data accountability or audit verification. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention is described below through specific examples shown in the embodiments. However, it should be understood that these descriptions are merely exemplary and not intended to limit the scope of the invention. Furthermore, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessarily obscuring the concepts of the invention.

[0020] Example

[0021] This invention relates to a method for recording the execution process log of a multi-dimensional auditing intelligent agent, which specifically includes the following steps:

[0022] S1, the audit system audits the uploaded document data and audits all audit points of the document according to the business rules for document audit;

[0023] S2, after each audit point of the document is completed, the audit system outputs the audit result for each audit point. The audit system outputs the final audit result based on all the audit results of the document, and at the same time outputs the text log of all audit points of the document and stores the text log in the database;

[0024] The review process is composed of a series of review points, following business rules. Each review point verifies one or more dimensions of data in the current document. All review points combined complete the verification of the document's completeness, rationality, consistency, and other aspects. Upon completion of each review point, a very detailed text log is output, including the review result, reason for review, review flow, physical processing method, start time, end time, and timeliness of the review. Each text log entry during the execution of a single review point includes detailed timestamps and is recorded in the database.

[0025] S3, When an audit point fails, the audit system stores the audit site data of that audit point in the database;

[0026] When an audit point fails, in addition to recording the aforementioned text information, the system will also preserve and record the audit scene at that audit point in the database. The audit scene refers to all types of original data involved at that audit point, including but not limited to the original bill of lading information, attachment information, ledger information, etc., that caused the audit failure. This original data will help auditors quickly locate the problem and make a quick judgment during manual intervention.

[0027] S4. If the audit point fails, a second manual audit is conducted. The second audit is conducted based on the text log and audit site data corresponding to the audit point, and the final audit result is output.

[0028] The final review result of this document will be determined by the priority of the results of each failed review point. The default priority in the system is rejection > transfer to manual review > communication. Therefore, if multiple review points fail, the overall review result of the document will be determined according to the default priority. That is, if there is a rejection review point, the final status of the document will be rejection.

[0029] In actual business operations, the same order may be reviewed multiple times in the system. In this case, the system will record the review logs of each review point and the overall review result of the order each time it enters the system, based on the different times the order enters the system. At the same time, it will retain the original data on site when each review fails.

[0030] The text log of each audit point includes the audit result, audit reason, audit flow, physical processing method, audit start time, audit end time, and audit timeliness; when each audit point is audited, the output text log also includes the detailed time of text log generation.

[0031] The audit results output by each audit point include audit passed and audit failed. Audit failed includes rejection, transfer to human assistant, and communication. When all audit results output by the audit points are audit passed during document audit, the document's audit result is audit passed. When an audit failed result appears in the audit point output during document audit, the document's audit result is audit failed.

[0032] After the document's audit result is output as a failure, the specific audit failure result is output according to the priority of the failed audit point. The priority of audit failure is, in order, rejection, transfer to manual, and communication.

[0033] When the audit point fails, the audit site data of the audit point includes the original bill of lading information, attachment information and ledger information that caused the audit failure.

[0034] The manual secondary review and confirmation process is as follows: when a document review fails, the manual reviewer determines the reason for the failure of the review point based on the review reason and the data at the review site; and based on the reason for the failure of the review point, the manual reviewer selects the corresponding reason for the failure of the review point and modifies the review result accordingly, while recording the dimensional information of the manual operation.

[0035] The reasons for the failure of the audit point include AI parsing errors, unreasonable rule configuration, missing bill of lading attachments, redundant bill of lading attachments, and incorrect bill of lading information; the dimension information of the manual operation includes the time of the audit point modification, operation type, operation content, and operator information.

[0036] When manually intervening in the rules for failed reviews, a failure label is added, and the entire process is recorded.

[0037] For audit points that fail system review, a second confirmation will be conducted manually. Based on the audit reason and the original data on site, the manual reviewer will quickly determine the reason for the error at the audit point. When modifying the audit point result, the reason for the failure at that audit point will be selected. Common reasons for failure include: AI parsing error, unreasonable rule configuration, missing bill of lading attachments, redundant bill of lading attachments, and incorrect bill of lading information.

[0038] After selecting the corresponding reason, the result of the audit point can be modified. At this time, the system will automatically record the time of modification, operation type, operation content, operator and other information. It also supports two different display formats, list and multi-dimensional, to facilitate future accountability or audit verification.

[0039] Analysis of reasons for post-audit failure

[0040] After a large number of documents are reviewed each month, the above log information can be used to conduct in-depth analysis of the review rules; for example, which review points frequently make mistakes, and the percentage of errors at each review point can be identified, thereby enabling well-founded optimization measures for AI capabilities, rule settings, and the company's bill of lading system.

[0041] The technical solution of this invention solves the problems of insufficient process records, lack of on-site content, and inability to conduct comprehensive analysis of failure reasons after the audit process. By recording multi-dimensional logs within the system, saving on-site audit data, and using a tagging system for manual intervention in audit failure results, the entire audit process can be recorded, and post-audit statistical analysis can be supported.

[0042] Obtain the audit results of the audit documents, and based on the failure options in the audit results, trace and display the reasons for the audit failure and the original data of the audit failure. Based on the original audit data, provide specific reasons for the failure, make corresponding modifications, and then resubmit for audit.

[0043] 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 preferred embodiments, those skilled in the art should understand that modifications can still be made to the specific implementation of the invention or equivalent substitutions can be made to some technical features without departing from the spirit of the technical solutions of the present invention, and all such modifications and substitutions should be covered within the scope of the technical solutions claimed in the present invention.

Claims

1. A method for implementing a process log record based on a multi-dimensional audit agent, characterized in that, The method specifically comprises the following steps: S1, the auditing system audits the uploaded bill data, and audits all auditing points of the bill according to the business rules of bill auditing; S2, after each auditing point of the bill is audited, the auditing system outputs the auditing result of each auditing point, the auditing system outputs the final auditing result according to all auditing results of the bill, outputs the text log of all auditing points of the bill, and stores the text log in the database; S3, when the auditing point fails, the auditing system stores the auditing site data of the auditing point in the database; S4, after the auditing point fails, manual secondary auditing confirmation is performed, secondary auditing confirmation is performed according to the text log and the auditing site data corresponding to the auditing point, and the final auditing result is output. 2.The method of claim 1, wherein, The text log of the auditing point includes the auditing result, the auditing reason, the auditing flow direction, the physical processing mode, the auditing point start execution time, the auditing point end execution time, and the auditing time limit; when each auditing point is audited, the output text log further includes the detailed time of the text log generation. 3.The method of claim 1, wherein, The auditing result output by each auditing point includes auditing pass and auditing failure, the auditing failure includes rejection, manual transfer and communication; when the auditing points output the auditing result of auditing pass during the bill auditing, the auditing result output by the bill is auditing pass; when the result output by the auditing points during the bill auditing includes auditing failure, the auditing result output by the bill is auditing failure.

4. The method of claim 3, wherein the method further comprises: After the auditing result output of the bill fails, the specific auditing failure result is output according to the priority of the failed auditing point, and the priority of the auditing failure is rejection, manual transfer and communication in turn.

5. The method of claim 1, wherein the method further comprises: When the auditing point fails, the auditing site data of the auditing point includes the original bill information, the attachment information and the account information causing the auditing failure.

6. The method of claim 1, wherein the method further comprises: The manual secondary auditing confirmation specifically includes: when the bill auditing fails, the human being judges the auditing failure reason of the auditing point according to the auditing reason and the auditing site data of the auditing point; and according to the auditing failure reason of the auditing point, the human being selects the corresponding auditing failure reason of the auditing point, and modifies the auditing result correspondingly, while recording the dimension information of the human operation.

7. The method of claim 6, wherein the method further comprises: The auditing failure reason of the auditing point includes AI analysis error, unreasonable rule configuration, bill attachment missing, bill attachment redundancy and bill information filling error; the dimension information of the human operation includes the time of the auditing point modification, the operation type, the operation content and the operator information.