CN-121998596-A - Multi-mode internal control compliance intelligent audit method and system for complex document

CN121998596ACN 121998596 ACN121998596 ACN 121998596ACN-121998596-A

Abstract

The invention discloses a multi-mode internal control compliance intelligent auditing method and system for complex documents, which relate to the technical field of intelligent auditing and comprise the steps of acquiring a system text, an approval record, a business document, a contract attachment and an image scanning piece, carrying out structural analysis to obtain system semantic information, business behavior information and evidence information, constructing a standard approval path, comparing the standard approval path with the approval record to obtain an approval path comparison result, identifying a system version and a time tag, constructing a system version temporal semantic mapping model, matching a corresponding system version according to business behavior occurrence time, obtaining a system version temporal semantic mapping result, analyzing a system rule, identifying exception triggering conditions, constructing a system exception term semantic rule map, judging whether a business behavior is applicable to an exception rule, and obtaining an exception rule judgment result.

Inventors

XU JIAWEN
HUANG FENG
Li Shilue
WU ANLE
XIE QIN
FENG MINGXING

Assignees

湖南嘉创信息科技发展有限公司

Dates

Publication Date: 20260508
Application Date: 20260410

Claims (10)

1. A multi-mode internal control compliance intelligent audit method facing complex documents is characterized by comprising the following steps: Obtaining a system text, an approval record, a business document, a contract attachment and an image scanning piece, and carrying out structural analysis to obtain system semantic information, business behavior information and evidence information, constructing a standard approval path, and then comparing the standard approval path with the approval record to obtain an approval path comparison result; identifying a system version and a time tag, constructing a system version temporal semantic mapping model, and matching the corresponding system version according to the occurrence time of the business behavior to obtain a system version temporal semantic mapping result; analyzing the rule of the system, identifying exception triggering conditions, constructing a semantic rule map of exception clauses of the system, judging whether the business behavior is applicable to the exception rule, and acquiring an exception rule judgment result; Performing source identification, structure analysis and path generation tracking on the business bill, the contract attachment and the image scanning piece, constructing a multi-mode evidence chain, performing reliability quantification calculation on evidence nodes in the multi-mode evidence chain, and obtaining an evidence reliability quantification result; Constructing an evidence-rule coupling relation diagram based on the rule information and the evidence credibility quantification result, calculating rule application credibility of candidate rule, and resolving conflict rules by combining the rule hierarchy relation to obtain an effective rule set; Constructing a system applicable state transition diagram based on the approval path comparison result, the system version temporal semantic mapping result, the exception rule judgment result and the effective rule set, and determining a target system applicable path; And constructing an audit reasoning path relation diagram along a target system applicable path, generating an audit interpretation path, comprehensively analyzing according to the audit interpretation path and the evidence credibility quantification result, and outputting a compliance risk level and a corresponding audit interpretation result.
2. The multi-mode internal control compliance intelligent auditing method for the complex document according to claim 1 is characterized in that the obtaining of the approval path comparison result comprises the steps of extracting approval nodes and sequence relations among the nodes according to approval flow rules in a system text, constructing a standard approval path, extracting actual approval nodes and execution sequences thereof from an approval record to form an actual approval path, carrying out node matching and sequence consistency comparison on the actual approval path corresponding to the approval record and the standard approval path, identifying approval nodes which exist in the standard approval path but do not exist in the actual approval path, and taking the identified approval nodes as implicit missing nodes to generate an approval path comparison result.
3. The multi-modal internal control compliance intelligent auditing method for the complex document is characterized in that the acquisition of the system version temporal semantic mapping result comprises the steps of extracting system identifications, version identifications and time tag information from system texts, classifying the system texts of different versions according to the system identifications, constructing a system version temporal semantic mapping model based on the version identifications and the time tag information, wherein the system version temporal semantic mapping model is used for establishing a corresponding relation between each system version and the corresponding system semantic information and the application time interval of each system version, determining the application time interval of each system version according to the time tag corresponding to each system version, mapping the service behavior occurrence time to the corresponding application time interval, determining the system version corresponding to the service behavior information through the system version temporal semantic mapping model, and generating the system version temporal semantic mapping result.
4. A multi-mode internal control compliance intelligent auditing method for complex documents according to claim 1 is characterized in that the obtaining of exception rule judging results comprises the steps of carrying out rule semantic analysis on a system text, identifying condition contents in the system rule and corresponding exception triggering conditions, carrying out association storage on the system rule and the corresponding exception triggering conditions, constructing a system exception clause semantic rule map, wherein association relations between different system rules and the corresponding exception triggering conditions are respectively established, extracting feature information corresponding to the exception triggering conditions according to business behavior information, matching the feature information with the exception triggering conditions, determining whether the business behavior meets the exception triggering conditions, judging that the corresponding system rule is applicable to the exception rule when the exception triggering conditions are met, and generating exception rule judging results.
5. The multi-mode internal control compliance intelligent auditing method for the complex document according to claim 1 is characterized in that the evidence reliability quantifying result obtaining comprises the steps of dividing evidence information obtained through analysis of business documents, contract attachments and image scanning pieces into a plurality of evidence nodes, determining time sequence relations among the evidence nodes according to generation time information corresponding to the evidence nodes, determining source association relations among the evidence nodes according to source information of the evidence nodes, determining content corresponding relations among the evidence nodes according to content information of the evidence nodes, establishing association relations among the evidence nodes based on the time sequence relations, the source association relations and the content corresponding relations, constructing a multi-mode evidence chain, and performing reliability quantifying calculation on the evidence nodes based on the generation time information, the source information and the content consistency of the evidence nodes to obtain the evidence reliability quantifying result.
6. The multi-mode internal control compliance intelligent auditing method for the complex document is characterized in that the construction of the evidence-rule coupling relation diagram comprises the steps of extracting rule element information corresponding to each rule according to rule information, carrying out association matching on the rule element information and evidence nodes to determine evidence node sets on which each rule depends, establishing association relations between the rule and the evidence nodes based on the evidence node sets and corresponding evidence reliability quantification results to construct the evidence-rule coupling relation diagram, carrying out weighted calculation according to evidence reliability quantification results of each evidence node in the evidence node set corresponding to the rule, and carrying out correction by combining content consistency and time sequence relations among the evidence nodes to obtain rule applicability confidence.
7. The multi-mode internal control compliance intelligent auditing method for the complex document according to claim 1 is characterized in that the obtaining of the effective rule set comprises the steps of carrying out consistency comparison on the application results of candidate system rules, judging that conflict rules exist when the application results of different candidate system rules on the same business behavior are inconsistent, sorting the candidate system rules according to the rule application confidence of each candidate system rule, calculating evidence credibility distribution characteristics based on evidence credibility quantification results of evidence node sets corresponding to each candidate system rule, correcting rule application confidence according to the evidence credibility distribution characteristics, comprehensively comparing the corrected rule application confidence according to a system level relation, screening the candidate system rules with rule application confidence meeting preset conditions as the effective rules, and forming the effective rule set.
8. The multi-mode internal control compliance intelligent auditing method for the complex document is characterized in that the target system applicable path determination comprises the steps of combining approval completion states, system versions and rule applicable results to generate a plurality of candidate system applicable states, carrying out time sequence arrangement on the candidate system applicable states according to business behavior occurrence time, establishing state transition relations among the candidate system applicable states according to exception rule judging results, approval path comparison results, effective rule sets and evidence credibility quantification results to obtain a system applicable state transition diagram, carrying out constraint consistency check on each state transition relation in the system applicable state transition diagram, deleting state transition relations which do not meet constraint conditions, carrying out path traversal on the reserved state transition relations to obtain a plurality of candidate paths, and comparing the rule applicable credibility and the evidence credibility quantification results corresponding to the candidate paths to determine the target system applicable paths corresponding to business behavior information.
9. The multi-mode internal control compliance intelligent auditing method for complex documents according to claim 1 is characterized in that comprehensive analysis comprises the steps of carrying out association expansion on each system version, rule application result and corresponding evidence node in a path according to a state transition sequence along a target system application path, constructing an auditing reasoning path relation diagram, generating an auditing interpretation path, carrying out consistency assessment on each rule application result according to evidence credibility quantification results of the evidence node corresponding to each rule application result in the auditing interpretation path, correcting or eliminating rule application results which do not meet consistency conditions, carrying out compliance assessment on business behavior information based on the corrected rule application results, and outputting a compliance risk level and a corresponding auditing interpretation result.
10. A complex document oriented multi-modal internal control compliance intelligent auditing system, characterized in that it is used for implementing a complex document oriented multi-modal internal control compliance intelligent auditing method according to any one of claims 1 to 9, comprising: the path comparison module is used for acquiring a system text, an approval record, a business bill, a contract attachment and an image scanning piece, carrying out structural analysis to obtain system semantic information, business behavior information and evidence information, constructing a standard approval path, and then comparing the standard approval path with the approval record to acquire an approval path comparison result; The version mapping module is used for identifying the system version and the time tag, constructing a system version temporal semantic mapping model, matching the corresponding system version according to the occurrence time of the business behavior, and obtaining a system version temporal semantic mapping result; the exception judging module analyzes the rule of the system, identifies exception triggering conditions, constructs a semantic rule map of exception clauses of the system, judges whether the business behavior is applicable to the exception rule, and acquires an exception rule judging result; The credibility measuring module is used for carrying out source identification, structure analysis and path tracking generation on the business bill, the contract attachment and the image scanning piece, constructing a multi-mode evidence chain, carrying out credibility quantification calculation on evidence nodes in the multi-mode evidence chain and obtaining evidence credibility quantification results; The effective rule module is used for constructing an evidence-rule coupling relation diagram based on the rule information and the evidence credibility quantification result, calculating rule applicable confidence level of the candidate rule, and resolving conflict rules by combining the rule hierarchical relation to obtain an effective rule set; The system application module is used for constructing a system application state transition diagram based on the examination and approval path comparison result, the system version temporal semantic mapping result, the exception rule judgment result and the effective rule set and determining a target system application path; And the comprehensive audit module is used for constructing an audit reasoning path relation diagram along the target system applicable path, generating an audit interpretation path, carrying out comprehensive analysis according to the audit interpretation path and the evidence credibility quantification result, and outputting a compliance risk level and a corresponding audit interpretation result.

Description

Multi-mode internal control compliance intelligent audit method and system for complex document Technical Field The invention relates to the technical field of intelligent audit, in particular to a multi-mode internal control compliance intelligent audit method and system for complex documents. Background Along with the deep digital transformation of enterprises, the data environment faced by the internal control compliance audit has evolved from a single structured record to a complex scene of multi-mode and multi-source heterogeneous documents such as coverage system text, approval record, business document, contract attachment, image scanning piece and the like. The traditional auditing method relies on manual review, comparison and judgment of auditors, is low in efficiency, and is difficult to deal with deep logic problems such as hidden rule temporal change, exception clause recognition, evidence chain association and rule conflict resolution in mass documents. In recent years, some intelligent audit systems attempt to introduce a rules engine or knowledge base to translate institutional rules into executable decision logic in an effort to promote the level of audit automation. However, the prior art still has significant shortcomings in terms of relevance, dynamics, and uncertainty in processing multi-modal documents. Firstly, the rule of the system according to the audit has the conditions of parallel multi-version and staggered effective time intervals, but the existing system is difficult to automatically match the applicable system version based on the occurrence time of the business behavior, so that the rule is applicable and the temporal dislocation occurs. Secondly, exception terms in the system are generally expressed in a scattered way in a natural language form, and the system cannot accurately judge whether the business behavior belongs to an exception condition due to the lack of structural semantic modeling. More importantly, the lack of unified association and credibility quantification mechanism between the evidence information of different sources, such as documents, images and approval records, leads to broken evidence chains or incapability of evaluating evidence quality, and further influences the reliability of rule judgment results. When multiple rules produce conflicting conclusions about the same business behavior, existing systems also lack rule preference capabilities based on evidence credibility and system level. Disclosure of Invention The invention provides an intelligent audit method for realizing dynamic adaptation of a system version, accurate recognition of exception rules, unified modeling of a multi-modal evidence chain and intelligent resolution of rule conflicts under a complex document environment with heterogeneous multi-source, staggered temporal and rule conflicts. In order to achieve the above purpose, the technical scheme adopted by the invention is as follows: A multi-mode internal control compliance intelligent audit method facing complex documents comprises the following steps: Obtaining a system text, an approval record, a business document, a contract attachment and an image scanning piece, and carrying out structural analysis to obtain system semantic information, business behavior information and evidence information, constructing a standard approval path, and then comparing the standard approval path with the approval record to obtain an approval path comparison result; identifying a system version and a time tag, constructing a system version temporal semantic mapping model, and matching the corresponding system version according to the occurrence time of the business behavior to obtain a system version temporal semantic mapping result; analyzing the rule of the system, identifying exception triggering conditions, constructing a semantic rule map of exception clauses of the system, judging whether the business behavior is applicable to the exception rule, and acquiring an exception rule judgment result; Performing source identification, structure analysis and path generation tracking on the business bill, the contract attachment and the image scanning piece, constructing a multi-mode evidence chain, performing reliability quantification calculation on evidence nodes in the multi-mode evidence chain, and obtaining an evidence reliability quantification result; Constructing an evidence-rule coupling relation diagram based on the rule information and the evidence credibility quantification result, calculating rule application credibility of candidate rule, and resolving conflict rules by combining the rule hierarchy relation to obtain an effective rule set; Constructing a system applicable state transition diagram based on the approval path comparison result, the system version temporal semantic mapping result, the exception rule judgment result and the effective rule set, and determining a target system applicable path; And constructing an audit reas