A document compliance determination method based on violation tracing path scoring

Abstract

This invention discloses a document compliance determination method based on violation tracing path scoring, belonging to the field of data compliance review technology. The method includes: performing structured parsing of the document to be inspected, breaking down the document content into semantic units, and identifying the sensitive information entities contained therein; constructing a violation tracing path from the sensitive information entities to the relevant legal provisions; calculating a score for the violation tracing path, determining the violation risk level of the path based on the attributes of each node and edge in the path; matching the corresponding legal provisions based on the path scoring results, extracting the original text, provision number, applicable conditions, jurisdictional classification information, and penalty suggestions from a legal knowledge graph, and generating an interpretable causal chain description; and outputting a violation result report. This invention solves the problems of untraceable compliance basis, opaque violation determination, and lack of explanatory support for risk classification in existing document review technologies.

Description

Technical Field

[0001] This invention relates to the field of data compliance review technology, specifically a document compliance determination method based on violation tracing path scoring. Background Technology

[0002] With increasingly stringent requirements for data security and privacy protection, many industries are facing increasingly complex compliance auditing pressures in document management, content sharing, and cross-departmental information flow. For example, regulations such as GDPR, HIPAA, and the Personal Information Protection Act have imposed clear restrictions on the handling of sensitive information such as personal, health, and financial information in documents.

[0003] Current document compliance reviews primarily rely on rule-based or keyword-based identification methods, with some systems employing entity recognition and document classification technologies for sensitive information detection. However, these methods typically only provide a judgment on the existence of risky content, making it difficult to clearly explain the basis for violations, especially failing to trace the source to specific legal provisions or behavioral trigger chains, resulting in a lack of transparency and auditability.

[0004] Furthermore, different regulations have multiple definitions and processing requirements for similar data, making it difficult for traditional methods to perform fine-grained matching and personalized risk assessment. Therefore, there is an urgent need for a violation identification mechanism that supports "traceability, scoring, and explainability" to improve the accuracy and reliability of document compliance determination. Summary of the Invention

[0005] This invention addresses the problems existing in current document review technologies, such as untraceable compliance evidence, opaque violation determination, and lack of explanatory support for risk classification, by providing a document compliance determination method based on violation tracing path scoring.

[0006] The present invention provides a document compliance determination method based on violation tracing path scoring, and the technical solution adopted to solve the above-mentioned technical problems is as follows:

[0007] A document compliance determination method based on violation tracing path scoring includes the following steps:

[0008] S1. Perform structured parsing on the document to be inspected, break down the document content into semantic units, and identify the sensitive information entities contained therein to form the basic data layer for compliance auditing;

[0009] S2. Construct a violation tracing path from the sensitive information entity to the legal clause, wherein the violation tracing path includes content fragments, entity types, rule triggering conditions, and legal clause nodes;

[0010] S3. Calculate a score for the violation tracing path and determine the violation risk level of the path based on the attributes of each node and edge in the path;

[0011] S4. Based on the scoring results of the violation tracing path, match the corresponding legal provisions, extract the original text, clause number, applicable conditions, jurisdiction classification information and penalty suggestions of the legal provisions from the legal knowledge graph, provide clear legal basis for violation identification, and generate an interpretable causal chain description.

[0012] S5. Output a violation result report, which includes the location information of the violation content, the source tracing path structure, the rating level, the matching legal basis, and the handling suggestions.

[0013] Optionally, step S1 specifically includes:

[0014] S1.1 Receive various documents to be tested, parse the documents to be tested, and complete content extraction and structured processing;

[0015] S1.2. Using a text segmentation algorithm, the parsed content of the document to be detected is divided into logically clear basic semantic units;

[0016] S1.3 Apply the entity recognition model to scan the text in the basic semantic unit, identify sensitive information entities, add a label to each identified entity, and record the basic semantic unit in which it is located.

[0017] S1.4. Mark the basic semantic unit set of sensitive information entities to form an "entity-location-type" mapping table, providing initial nodes for subsequent illegal path construction.

[0018] Preferably, the documents to be detected include three types: structured documents, semi-structured documents, and image documents;

[0019] a) For structured and semi-structured documents, directly parse the document format, extract the text content and formatting information, and output structured text data that retains the original layout.

[0020] b) For image-based documents, the OCR module is first called to perform text recognition on the image and generate editable text content. Then, the layout analysis technology is used to restore the document layout, distinguishing the main text, header, footer and table areas. Finally, the text data with coordinate positions and layout structure information are output.

[0021] Further optionally, step S2 specifically includes:

[0022] S2.1 Based on the recognition results of the document to be detected, extract the entities marked as sensitive information and the content segments in which they are located;

[0023] S2.2. Based on the sensitive entity type, retrieve the predefined set of regulations and rules, match the corresponding compliance rule conditions, and determine whether the rule triggering conditions are met by combining the context semantics.

[0024] S2.3. Based on the matched rule triggering conditions, query the predefined compliance knowledge graph, locate the corresponding regulatory clause node, parse the content of the regulatory clause, extract structured information, and verify the matching with the current violation scenario;

[0025] S2.4 Connect “content fragments → entity types → rule triggering conditions → legal clauses” in logical order to form a structured tracing path, supplement the contextual semantic features in the tracing path, and ensure that the path is semantically closed and logically reasonable.

[0026] Further optionally, step S3 specifically includes:

[0027] S3.1 For the completed violation tracing path, extract five attributes of nodes and edges in the path: content sensitivity of content fragments, entity confidence of entity type, rule matching degree between rule triggering conditions and path semantics, clause mandatoryness of legal clauses, and contextual coherence of the entity's context, and generate a vector containing multi-dimensional features.

[0028] S3.2. The feature vector is comprehensively calculated by a preset weighted scoring function, and confidence interval modeling is incorporated to handle fuzzy semantic scenarios. It also supports manual parameter adjustment or automatic optimization through reinforcement learning to achieve quantitative assessment of risk level.

[0029] S3.3 Based on the set risk threshold, the calculation results are divided into three levels: "high risk", "medium risk" and "low risk". Then, a detailed report containing risk scores, factor contribution analysis and regulatory references is generated. Confidence descriptions are added to fuzzy paths to assist decision-making.

[0030] Further optionally, step S4 specifically includes:

[0031] S4.1. Expand the search of the end node of the violation tracing path in the legal knowledge graph to obtain the legal content corresponding to the end node;

[0032] S4.2. Conduct in-depth analysis of the semantic information contained in the violation tracing path and the rule information of the legal clauses in the legal knowledge graph. Through detailed comparison, achieve intelligent alignment between the violation tracing path and the legal content, and accurately identify the legal clauses corresponding to the end node of the path.

[0033] S4.3 Extract detailed information related to the clause from the legal knowledge graph, including the original text of the legal clause, clause number, applicable conditions, jurisdictional classification information, and penalty suggestions, to provide a reference for subsequent handling of violations;

[0034] S4.4 When a violation tracing path corresponds to multiple legal provisions, multiple candidate provisions are output simultaneously, and the multiple candidate provisions are scored and ranked according to the path score and the degree of matching between the rule and the path.

[0035] S4.5. Using the established violation tracing path and the matched legal provisions, generate an interpretable causal chain description.

[0036] Further optionally, step S5 specifically includes:

[0037] S5.1 Collect risk scoring results, regulatory compliance information, violation tracing paths, and related compliance judgment information, organize and classify them to prepare for subsequent report generation;

[0038] S5.2. Package the collected and organized information according to the set structure;

[0039] S5.3. Based on the structured and encapsulated information, generate the specific content of the violation result report. The report includes the location information of the violation content, the source tracing path structure, the rating level, the matching legal basis, and the handling suggestions.

[0040] Preferably, step S5 is executed, and a violation result report is output in the following manner:

[0041] a) Display the report content directly on the human-machine interface;

[0042] b) Automatically generates PDF reports for easy saving and distribution;

[0043] c) Provide structured data interfaces to connect with audit systems, business process systems, risk control platforms, or automated processing engines to achieve data sharing and interaction.

[0044] The document compliance determination method based on violation tracing path scoring of the present invention has the following advantages compared with the prior art:

[0045] 1. This invention constructs a path for tracing violations from document content fragments to legal provisions. Based on multiple factors such as the content characteristics of the path, the degree of entity matching, the complexity of rule conditions, and the semantic relevance of the context, it scores the document content to determine whether it constitutes a violation and the level of violation. It also generates legal tracing information and handling suggestions, thereby achieving risk highlighting, violation location, and interpretability compliance review in the process of handling sensitive information in documents.

[0046] 2. This invention improves the accuracy and semantic depth of violation identification through structured source tracing modeling of violations; it supports quantitative scoring and classification of the risks of violating content, meeting the needs of different organizations, industries, or regions for refined compliance management and controllable risks; it can clearly output information such as "which rule was triggered by the violation, which regulation applies, and why it constitutes a violation," supporting regulation source tracing at the clause level, providing legal support and process transparency for regulatory audits, compliance appeals, and manual review; it can automatically complete the compliance closed-loop processing from identification, judgment, interpretation to recommendations, supporting content screening, document blocking, or rewriting processes without human intervention, reducing enterprise compliance costs, and improving organizational risk response efficiency.

[0047] 3. This invention realizes a closed-loop control of the entire process from document semantic understanding to intelligent comparison of regulations, and from violation judgment to interpretation output. It is particularly suitable for practical application scenarios with high requirements for content security and compliance, complex text structure, and flexible application of regulations. Attached Figure Description

[0048] Appendix Figure 1 This is a flowchart of the method according to Embodiment 1 of the present invention. Detailed Implementation

[0049] To make the technical solution, the technical problem solved, and the technical effect of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with specific embodiments.

[0050] Example 1:

[0051] Reference Appendix Figure 1 This embodiment proposes a document compliance determination method based on violation tracing path scoring, which includes the following steps:

[0052] S1. Perform structured parsing on the document to be inspected, break down the document content into semantic units, and identify the sensitive information entities contained therein to form the basic data layer for compliance auditing, specifically including:

[0053] S1.1 Receive various documents to be tested, parse the documents to be tested, and complete content extraction and structured processing.

[0054] S1.2. Using a text segmentation algorithm, the parsed document content is divided into logically clear basic semantic units, such as paragraphs, sentences, table units, and heading blocks.

[0055] S1.3 Apply the entity recognition model to scan the text in the basic semantic unit, identify sensitive information entities such as name, address, contact information, identity information, and health data, add a label to each identified entity, and record the basic semantic unit in which it is located.

[0056] S1.4. Mark the basic semantic unit set of sensitive information entities to form an "entity-location-type" mapping table, providing initial nodes for subsequent illegal path construction.

[0057] It should be added that the documents to be detected are not limited to three types: structured documents (such as Word and PDF), semi-structured documents (such as HTML and email body), and image documents (such as scanned documents);

[0058] a) For structured and semi-structured documents, directly parse the document format, extract the text content and formatting information, and output structured text data that retains the original layout.

[0059] b) For image-based documents, the OCR module is first called to perform text recognition on the image and generate editable text content. Then, the layout analysis technology is used to restore the document layout, distinguishing the main text, header, footer and table areas. Finally, the text data with coordinate positions and layout structure information are output.

[0060] S2. Construct a violation tracing path from the sensitive information entity to the regulatory clauses. The violation tracing path includes content fragments, entity types, rule triggering conditions, and regulatory clause nodes, specifically including:

[0061] S2.1 Based on the identification results of the document to be detected, extract entities marked as sensitive information (such as names, ID numbers, health data, financial account information, etc., which are the starting point for constructing the violation tracing path) and their corresponding content segments (such as sentences, paragraphs, etc.); for example, if "Zhang San's ID number is 123456789012345678" in the document is identified as containing sensitive information, then "Zhang San" and "123456789012345678" are the sensitive information entities, and "Zhang San's ID number is 123456789012345678" is the corresponding content segment.

[0062] S2.2. Based on the sensitive entity type, retrieve the predefined set of regulations and rules, match the corresponding compliance rule conditions, and determine whether the rule triggering conditions are met by combining the context semantics.

[0063] In this step, the predefined set of regulatory rules is a series of rules related to regulations stored in the system. These rules stipulate the compliance requirements for different types of sensitive information in different scenarios. By searching the set of regulatory rules according to the type of sensitive entity, the corresponding compliance rule conditions can be found. For example, if the sensitive entity type is "personal identification information", the rule "processing personal identification information requires explicit authorization from the user" may be found.

[0064] Contextual semantics includes the purpose of using sensitive information, the method of processing it, and the subjects involved. For example, in a document, it states that "Zhang San's ID number is 123456789012345678, used for employee registration, and authorized by Zhang San." The context indicates that although sensitive information exists, it meets the triggering condition of the rule "processing personal identity information requires explicit user authorization." However, if no authorization information is mentioned, this condition may not be met.

[0065] S2.3. Based on the matched rule triggering conditions, query the predefined compliance knowledge graph, locate the corresponding regulatory clause node, parse the content of the regulatory clause, extract structured information, and verify the matching with the current violation scenario.

[0066] A predefined compliance knowledge graph is a structured way of representing knowledge. It organizes information such as regulatory clauses, rule conditions, and entity types in the form of a graph, which facilitates querying and reasoning.

[0067] For example, if the rule is triggered by "processing personal identity information without the user's explicit authorization," the relevant legal provisions in the Personal Information Protection Law should be located. The content of these legal provisions should be analyzed to extract structured information, such as the scope of application, liability provisions, and penalties. The extracted structured information should be compared and verified with the current violation scenario to determine whether the legal provisions are truly applicable to the current scenario. For example, if the scope of application of the legal provisions is "enterprises processing personal information in commercial activities," while the current scenario is "non-profit organizations processing personal information in public welfare activities," then it is necessary to further determine whether the legal provisions can fully cover the current scenario.

[0068] S2.4 Connect “content fragment → entity type → rule triggering condition → legal clause” in logical order to form a structured tracing path. For example, “Zhang San’s ID number is 123456789012345678 (content fragment) → personal identity information (entity type) → processing personal identity information without the user’s explicit authorization (rule triggering condition) → Article XX of the Personal Information Protection Law (legal clause)”;

[0069] Supplement the tracing path with contextual semantic features (such as the purpose of using sensitive information, processing methods, and involved entities) to ensure the path is semantically closed and logically sound. For example, adding contextual semantics such as "used for employee onboarding registration but without authorization" to the above path makes the logic of the entire path clearer.

[0070] S3. Calculate a score for the violation tracing path. Based on the attributes of each node (content fragment, entity type, rule triggering condition, legal clause) and edge (semantic relationship between nodes) in the path, determine the violation risk level of the path, specifically including:

[0071] S3.1 For the completed violation tracing path, extract five attributes of nodes and edges in the path: content sensitivity of content fragments, entity confidence of entity type, rule matching degree between rule triggering conditions and path semantics, clause mandatoryness of legal clauses, and contextual synergy of the entity's context, and generate a vector containing multi-dimensional features.

[0072] Content sensitivity: Assigned based on the type of sensitive information and the risk of leakage, such as ID number (10 points), name (6 points), health data (9 points), and dynamically adjusted in combination with the form of content exposure (plaintext transmission +3 points, encrypted storage 0 points).

[0073] Entity confidence score: assesses the accuracy of entity recognition, and is a probability value output by the OCR / NER model (e.g., 0.92 confidence score for ID card number recognition and 0.75 confidence score for fuzzy name recognition).

[0074] Rule matching degree: Judge the degree of fit between the rule triggering condition and the path semantics. A complete match of mandatory rules (such as "cross-border transmission requires security assessment") scores 10 points, while partial matches or exception scenarios score 4-7 points.

[0075] Mandatory nature of clauses: Based on the level of legal force, such as laws (Personal Information Protection Law) with a mandatory level of 10 points, administrative regulations (such as the Data Security Management Regulations) with 8 points, and industry standards with 5 points.

[0076] Contextual coherence: Analyze the compliance risks of the entity's usage scenarios, such as "cross-border transfer of health data" with a coherence score of 10 (high-risk scenario) and "internal archive encrypted storage" with a coherence score of 3 (low-risk scenario).

[0077] After quantifying the five attributes into numerical values, they are combined into a multidimensional feature vector (e.g., [9,0.92,8,10,10]) to provide a data foundation for subsequent scoring.

[0078] S3.2. The feature vector is comprehensively calculated by a preset weighted scoring function, and confidence interval modeling is incorporated to handle fuzzy semantic scenarios. It also supports manual parameter adjustment or automatic optimization through reinforcement learning to achieve quantitative assessment of risk level.

[0079] In this step, the weighting logic is as follows: weights are allocated according to the business compliance priorities, for example, content sensitivity (40%) > mandatory terms (30%) > rule matching degree (20%) > contextual consistency (8%) > entity confidence degree (2%). Example formula: Risk score = 0.4 × sensitivity + 0.3 × mandatory terms + 0.2 × rule matching degree + 0.08 × contextual consistency degree + 0.02 × entity confidence degree.

[0080] S3.3 Based on the set risk thresholds (e.g., high risk ≥ 80 points, medium risk 60-79 points, low risk < 60 points), the calculation results are divided into three levels: "high risk", "medium risk" and "low risk". Then, a detailed report containing risk scores, factor contribution analysis and regulatory references is generated, and confidence statements are added to fuzzy paths to assist decision-making.

[0081] S4. Based on the scoring results of the violation tracing path, match the corresponding legal provisions, extract the original text, clause number, applicable conditions, jurisdiction classification information and penalty suggestions of the legal provisions from the legal knowledge graph, provide clear legal basis for violation identification, and generate an interpretable causal chain description.

[0082] This step specifically includes:

[0083] S4.1. Expand the search of the end node of the violation tracing path in the legal knowledge graph to obtain the legal content corresponding to the end node;

[0084] S4.2. Conduct in-depth analysis of the semantic information contained in the violation tracing path and the rule information of the legal clauses in the legal knowledge graph. Through detailed comparison, achieve intelligent alignment between the violation tracing path and the legal content, and accurately identify the legal clauses corresponding to the end node of the path.

[0085] S4.3 Extract detailed information related to the clause from the legal knowledge graph, including the original text of the legal clause, clause number, applicable conditions, jurisdictional classification information, and penalty suggestions, to provide a reference for subsequent handling of violations;

[0086] S4.4 When a violation tracing path corresponds to multiple legal provisions, multiple candidate provisions are output simultaneously, and the multiple candidate provisions are scored and ranked according to the path score and the degree of matching between the rule and the path.

[0087] S4.5. Using the established violation tracing path and the matched legal provisions, generate an interpretable causal chain description.

[0088] S5. Output a violation result report, specifically including:

[0089] S5.1 Collect risk scoring results, regulatory matching information, violation tracing paths, and related compliance judgment information. This includes the specific location of the violation content (such as page number, paragraph, specific sentence), the attributes of each node and edge in the path (used to generate the tracing path structure), risk scoring level, matching regulatory clause number, original text summary, and corresponding handling suggestions, etc.; organize and classify the collected information to prepare for subsequent report generation.

[0090] S5.2. Package the collected and organized information according to the set structure;

[0091] S5.3. Based on the structured and encapsulated information, generate the specific content of the violation result report. The report includes the location information of the violation content, the source tracing path structure, the rating level, the matching legal basis, and the handling suggestions.

[0092] Preferably, step S5 is executed, and a violation result report is output in the following manner:

[0093] a) Display the report content directly on the human-machine interface;

[0094] b) Automatically generates PDF reports for easy saving and distribution;

[0095] c) Provide structured data interfaces to connect with audit systems, business process systems, risk control platforms, or automated processing engines to achieve data sharing and interaction.

[0096] In summary, the document compliance determination method based on violation tracing path scoring of the present invention can improve the interpretability and accuracy of compliance judgment, and is particularly suitable for scenarios with high data compliance requirements such as enterprise document management, cross-border data review, and medical and financial services.

[0097] The above specific examples illustrate the principles and implementation methods of the present invention in detail. These embodiments are merely for the purpose of helping to understand the core technical content of the present invention. Based on the above specific embodiments of the present invention, any improvements and modifications made to the present invention by those skilled in the art without departing from the principles of the present invention should fall within the patent protection scope of the present invention.

Claims

1. A document compliance determination method based on violation tracing path scoring, characterized in that, Includes the following steps: S1. Perform structured parsing on the document to be inspected, break down the document content into semantic units, and identify the sensitive information entities contained therein to form the basic data layer for compliance auditing; S2. Construct a violation tracing path from the sensitive information entity to the regulatory clauses, wherein the violation tracing path includes content fragments, entity types, rule triggering conditions, and regulatory clause nodes; specifically including: S2.1 Based on the recognition results of the document to be detected, extract the entities marked as sensitive information and the content segments in which they are located; S2.

2. Based on the sensitive entity type, retrieve the predefined set of regulations and rules, match the corresponding compliance rule conditions, and determine whether the rule triggering conditions are met by combining the context semantics. S2.

3. Based on the matched rule triggering conditions, query the predefined compliance knowledge graph, locate the corresponding regulatory clause node, parse the content of the regulatory clause, extract structured information, and verify the matching with the current violation scenario; S2.4 Connect "content fragments → entity types → rule triggering conditions → legal clauses" in logical order to form a structured tracing path, supplement the contextual semantic features in the tracing path, and ensure that the path is semantically closed and logically reasonable; S3. Calculate a score for the violation tracing path, and determine the violation risk level of the path based on the attributes of each node and edge in the path; specifically including: S3.1 For the completed violation tracing path, extract five attributes of nodes and edges in the path: content sensitivity of content fragments, entity confidence of entity type, rule matching degree between rule triggering conditions and path semantics, clause mandatoryness of legal clauses, and contextual coherence of the entity's context, and generate a vector containing multi-dimensional features. S3.

2. The feature vector is comprehensively calculated by a preset weighted scoring function, and confidence interval modeling is incorporated to handle fuzzy semantic scenarios. It also supports manual parameter adjustment or automatic optimization through reinforcement learning to achieve quantitative assessment of risk level. S3.3 Based on the set risk threshold, the calculation results are divided into three levels: "high risk", "medium risk" and "low risk". Then, a detailed report containing risk scores, factor contribution analysis and regulatory references is generated. Confidence descriptions are added to fuzzy paths to assist decision-making. S4. Based on the scoring results of the violation tracing path, match the corresponding legal provisions, extract the original text, clause number, applicable conditions, jurisdiction classification information and penalty suggestions of the legal provisions from the legal knowledge graph, provide clear legal basis for violation identification, and generate an interpretable causal chain description. S5. Output a violation result report, which includes the location information of the violation content, the source tracing path structure, the rating level, the matching legal basis, and the handling suggestions.

2. The document compliance determination method based on violation tracing path scoring according to claim 1, characterized in that, Step S1 specifically includes: S1.1 Receive various documents to be tested, parse the documents to be tested, and complete content extraction and structured processing; S1.

2. Using a text segmentation algorithm, the parsed content of the document to be detected is divided into logically clear basic semantic units; S1.3 Apply the entity recognition model to scan the text in the basic semantic unit, identify sensitive information entities, add a label to each identified entity, and record the basic semantic unit in which it is located. S1.

4. Mark the basic semantic unit set of sensitive information entities to form an "entity-location-type" mapping table, providing initial nodes for subsequent construction of illegal paths.

3. The document compliance determination method based on violation tracing path scoring according to claim 2, characterized in that, The documents to be detected include three types: structured documents, semi-structured documents, and image documents. a) For structured and semi-structured documents, directly parse the document format, extract the text content and formatting information, and output structured text data that retains the original layout. b) For image-based documents, the OCR module is first called to perform text recognition on the image to generate editable text content. Then, layout analysis technology is used to restore the document layout, distinguishing the main text, header, footer, and table areas. Finally, text data with coordinate positions and layout structure information are output.

4. The document compliance determination method based on violation tracing path scoring according to claim 2, characterized in that, Step S4 specifically includes: S4.

1. Expand the search of the end node of the violation tracing path in the legal knowledge graph to obtain the legal content corresponding to the end node; S4.

2. Conduct in-depth analysis of the semantic information contained in the violation tracing path and the rule information of the legal clauses in the legal knowledge graph. Through detailed comparison, achieve intelligent alignment between the violation tracing path and the legal content, and accurately identify the legal clauses corresponding to the end node of the path. S4.3 Extract detailed information related to the clause from the legal knowledge graph, including the original text of the legal clause, clause number, applicable conditions, jurisdictional classification information, and penalty suggestions, to provide a reference for subsequent handling of violations; S4.4 When a violation tracing path corresponds to multiple legal provisions, multiple candidate provisions are output simultaneously, and the multiple candidate provisions are scored and ranked according to the path score and the degree of matching between the rule and the path. S4.

5. Using the established violation tracing path and the matched legal provisions, generate an interpretable causal chain description.

5. The document compliance determination method based on violation tracing path scoring according to claim 4, characterized in that, Step S5 specifically includes: S5.1 Collect risk scoring results, regulatory compliance information, violation tracing paths, and related compliance judgment information, organize and classify them to prepare for subsequent report generation; S5.

2. Package the collected and organized information according to the set structure; S5.

3. Based on the structured and encapsulated information, generate the specific content of the violation result report. The report includes the location information of the violation content, the source tracing path structure, the scoring level, the matching legal basis, and the handling suggestions.

6. The document compliance determination method based on violation tracing path scoring according to claim 5, characterized in that, Execute step S5 and output the violation result report in the following manner: a) Display the report content directly on the human-machine interface; b) Automatically generates PDF reports for easy saving and distribution; c) Provide structured data interfaces to connect with audit systems, business process systems, risk control platforms, or automated processing engines to achieve data sharing and interaction.

Images