Contract risk auditing method and device

Abstract

The invention relates to the field of artificial intelligence, and particularly provides a contract risk auditing method and device, and the method comprises the following steps: S1, splitting a contract content paragraph, and decomposing a contract text into a plurality of independent paragraph units with a logic relation; s2, converting the contract text into a structured data unit, and performing text analysis and risk factor extraction; s3, table textualization: converting table data in the contract into a plain text format which can be processed by a large model; s4, constructing a list table to systematically organize key information in the contract; s5, performing text semantic analysis to deeply understand the contract text; s6, element checking: carefully reviewing and verifying key elements in the contract text; and S7, risk information summarization: summarizing and analyzing potential risks in the contract text. Compared with the prior art, the method has the advantages that potential risks in the text can be recognized and predicted through learning of a large amount of text data, and therefore the efficiency and accuracy of contract risk auditing are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and specifically provides a contract risk auditing method and device. Background Art

[0002] In today's business environment, contracts, as important legal documents that regulate transaction behaviors and clarify rights and obligations, their risk management is crucial. However, traditional contract risk auditing methods mainly rely on manual review, which is not only inefficient but also easily affected by the experience and subjective judgment of auditors, making it difficult to ensure the accuracy and comprehensiveness of risk identification. Against this background, with the rapid development of artificial intelligence technology, especially the maturity of natural language processing (NLP) and machine learning technology, it provides a new technical path for contract risk auditing.

[0003] Traditional contract risk auditing means mainly include: manual review, where auditors identify potential risks by reading contract terms item by item and relying on professional knowledge and experience.

[0004] This method takes a long time and is prone to missing risk points; rule engines, which identify specific risks in contracts through predefined rules, but this method has poor flexibility and is difficult to handle complex and changeable contract terms; traditional text analysis tools, where early text analysis tools were mainly based on keyword search and pattern matching, lacking in-depth semantic understanding of contract texts, so the accuracy of risk identification is limited; database technology, which uses databases to store and manage contract information, but is usually limited to the processing of structured data and has insufficient processing capabilities for unstructured information in contract texts. Summary of the Invention

[0005] The present invention aims at the deficiencies of the above-mentioned existing technologies and provides a contract risk auditing method with strong practicability.

[0006] A further technical task of the present invention is to provide a contract risk auditing device with reasonable design, safety and applicability.

[0007] The technical solution adopted by the present invention to solve its technical problems is:

[0008] A contract risk auditing method has the following steps:

[0009] S1. Splitting of contract content paragraphs, decomposing the contract text into several independent and logically related paragraph units;

[0010] S2. The contract text is converted into structured data units for text analysis and risk factor extraction;

[0011] S3. Tabular text conversion, converting the tabular data in the contract into a pure text format that can be processed by a large model;

[0012] S4. List Table Construction: Systematically organize the key information in the contract to form a structured list table.

[0013] S5. Text Semantic Analysis: Through in-depth understanding of the contract text, identify potential risks and meanings in the contract terms.

[0014] S6. Element Verification: Conduct a detailed review and verification of the key elements in the contract text to discover and correct potential errors and risks.

[0015] S7. Risk Information Summary: Summarize and analyze the potential risks in the contract text to provide decision-making support for contract risk management.

[0016] Furthermore, in step S1, it includes:

[0017] S1-1. Preprocessing: First, preprocess the contract text. In addition, perform sentence splitting on the text to split continuous text into independent sentences.

[0018] S1-2. Title Recognition: Use the named entity recognition technology in NLP to identify the titles and subtitles in the contract.

[0019] S1-3. Paragraph Boundary Determination: After identifying the titles, determine the start and end boundaries of paragraphs based on the positions of the titles.

[0020] S1-4. Content Classification: Classify the identified paragraph content according to the preset paragraph categories.

[0021] S1-5. Recursive Subdivision: For some paragraphs containing multiple sub-clauses, adopt the method of recursive subdivision to further split the paragraphs into smaller units until the predetermined subdivision degree is reached.

[0022] S1-6. Logical Verification: After the splitting is completed, conduct logical verification.

[0023] S1-7. Result Output: Finally, output the split paragraphs in the form of structured data, with each paragraph carrying corresponding tags and serial numbers.

[0024] Furthermore, in step S3, it includes:

[0025] S3-1. Table Recognition: First, locate and recognize the table areas in the contract document through image recognition technology. Use computer vision algorithms to analyze the contract scanned image to distinguish text and table content.

[0026] S3-2. Table Structure Parsing: Parse the structure of the recognized table.

[0027] S3-3. Data extraction: Extract text data from the parsed table. Use optical character recognition technology to convert the image text in the table into an editable text format. For parts that are difficult to recognize by OCR, use a deep learning model for character recognition and correction;

[0028] S3-4. Format conversion: Convert the extracted table data into a unified text format. During the conversion process, preserve the original structural information of the table data;

[0029] S3-5. Data verification: Verify the converted text data;

[0030] S3-6. Text integration: Integrate the data after tabular textification with the contract body text to form a complete contract text.

[0031] Furthermore, in step S4, it includes:

[0032] S4-1. Information extraction: Extract the core elements in the contract based on the contract text;

[0033] S4-2. Classify the extracted information according to the nature and function of the contract terms;

[0034] S4-3. Table design: Design the framework structure of the checklist table according to the extracted and classified information;

[0035] S4-4. Data filling: Fill the extracted classified information into the corresponding positions of the checklist table;

[0036] S4-5. Verification and review: Verify the filled checklist table to ensure that all the entered data is consistent with the original contract text, without omission or error;

[0037] S4-6. Format optimization: Optimize the format of the checklist table so that it meets the audit requirements and is convenient for reading and analysis.

[0038] Furthermore, in step S5, it includes:

[0039] S5-1. Word segmentation and part-of-speech tagging: First, perform word segmentation on the contract text, split the continuous text into independent lexical units, and perform part-of-speech tagging on each word;

[0040] S5-2. Syntactic analysis: Construct the syntactic structure tree of the sentence by performing syntactic analysis on the contract text;

[0041] S5-3. Semantic role labeling: Identify the semantic roles in the sentence;

[0042] S5-4. Sentiment analysis: Perform sentiment analysis on the contract text to judge the sentiment tendency of the clause expression;

[0043] S5-5, Entity Recognition and Relationship Extraction: Using named entity recognition technology to identify key entities in the contract and extract the relationships between entities.

[0044] S5-6, Logical Relationship Analysis: Analyze the logical relationships between contract clauses to reveal the logical structure and dependency relationships between clauses.

[0045] S5-7, Risk Point Identification: Based on the above analysis results, identify potential risk points in the contract.

[0046] Furthermore, in step S6, it includes:

[0047] S6-1, Comparison Analysis: Compare the key elements in the contract text with the data in the list table to check whether they are consistent.

[0048] S6-2, Rule Verification: Verify the contract elements according to the preset rules and standards; during the verification process, if elements that do not conform to the rules are found, further investigation and analysis are required.

[0049] S6-3, Model Verification: Use a large model to perform deep learning on the contract elements to identify potential risks.

[0050] S6-4, Expert Review: On the basis of model verification, introduce an expert review link. Experts, according to their professional knowledge and experience, conduct further review of the contract elements.

[0051] S6-5, Feedback and Correction: According to the results of comparison analysis, rule verification, model verification, and expert review, feedback the discovered problems and correct the relevant elements in the contract text.

[0052] S6-6, Verification Record: Record the problems discovered during the verification process, the handling measures, and the correction results to form a verification record form.

[0053] Furthermore, in step S7, it includes:

[0054] S7-1, Risk Point Summary: Based on the results of text semantic analysis and element verification, summarize the potential risk points in the contract text.

[0055] S7-2, Risk Level Assessment: Assess the risk level of the summarized risk points and classify them into different levels according to the severity and impact scope of the risk points.

[0056] S7-3, Risk Cause Analysis: Conduct in-depth analysis of the risk points to find out the causes of the risks.

[0057] S7-4, Risk Response Strategy: For the identified risk points, propose corresponding risk response strategies.

[0058] S7-5. Risk information reporting, summarizing risk information, assessing risk levels, analyzing risk causes, and formulating risk response strategies to form a complete risk information report;

[0059] S7-6. Report approval and release, submitting the risk information report to the relevant department for approval, and releasing the risk information report after approval.

[0060] A contract risk audit device includes: at least one memory and at least one processor;

[0061] The at least one memory is used to store machine-readable programs;

[0062] The at least one processor is used to call the machine-readable program to execute a contract risk audit method.

[0063] Compared with the prior art, a contract risk audit method and device of the present invention have the following prominent beneficial effects:

[0064] By automatically processing contract texts, this method greatly reduces the workload of auditors and improves audit efficiency.

[0065] The large model has powerful semantic analysis capabilities, can more accurately identify contract risks, and reduces audit deviations caused by subjective judgments. It is applicable to various contract risk audits and has high practical value.

[0066] This method applies artificial intelligence technology to contract risk audits, providing new ideas for the innovative development of the audit industry. Through comprehensive analysis of texts, this method can identify various types of risks, ensuring the comprehensiveness of audit results. By constructing a checklist table, systematic collation of contract information is achieved, providing a clear data view for risk audits. By combining automation and manual review, information entry errors are reduced and data accuracy is improved.

[0067] As an intermediate product in the audit process, the checklist table facilitates information sharing and collaboration within the audit team. By verifying contract elements according to preset rules and standards, compliance with legal and regulatory requirements is ensured, reducing compliance risks.

[0068] In summary, this method has obvious advantages in the field of contract risk audits, can effectively improve the efficiency and accuracy of audit work, and provides strong technical support for contract risk management. Brief Description of the Drawings

[0069] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0070] Appendix Figure 1 is a schematic flowchart of a contract risk auditing method;

[0071] Appendix Figure 2 is a schematic diagram of the splitting of contract content paragraphs in a contract risk auditing method;

[0072] Appendix Figure 3 is a schematic diagram of the tabular text conversion in a contract risk auditing method;

[0073] Appendix Figure 4 is a schematic diagram of the construction of a list table in a contract risk auditing method;

[0074] Appendix Figure 5 is a schematic diagram of the semantic analysis of text in a contract risk auditing method;

[0075] Appendix Figure 6 is a schematic diagram of the element verification in a contract risk auditing method;

[0076] Appendix Figure 7 is a schematic diagram of the summary of risk information in a contract risk auditing method. Detailed implementation manners

[0077] To enable those in the technical field to better understand the solutions of the present invention, the following further elaborates on the present invention in combination with specific implementation manners. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0078] The following gives a best embodiment:

[0079] As Figure 1 shown, a contract risk auditing method in this embodiment has the following steps:

[0080] S1. Splitting of contract content paragraphs, decomposing the contract text into several independent and logically related paragraph units;

[0081] As Figure 2 shown, the structured decomposition of the contract text is achieved through the following steps:

[0082] S1-1, Preprocessing: First, preprocess the contract text, including removing irrelevant characters, unifying fonts and formats, correcting obvious typesetting errors, etc., to ensure the accuracy of text analysis. In addition, perform sentence segmentation on the text, splitting continuous text into independent sentences to lay the foundation for paragraph recognition.

[0083] S1-2, Title Recognition: Use the named entity recognition (NER) technology in NLP to identify the titles and subtitles in the contract. These titles are usually the signs at the beginning of paragraphs. By identifying keywords (such as "Article 1", "Chapter 2", etc.) and format features (such as bold, centered, etc.), the contract text can be accurately segmented into different paragraphs.

[0084] S1-3, Paragraph Boundary Determination: After identifying the titles, determine the start and end boundaries of paragraphs according to the positions of the titles. This step may involve syntactic analysis of sentences, dividing paragraphs by judging the end signs of sentences (such as full stops, semicolons, etc.).

[0085] S1-4, Content Classification: Classify the recognized paragraph content according to the preset paragraph categories (such as contract content, service requirements, etc.). This step may require the use of text classification algorithms, such as support vector machines (SVM), random forests, or deep learning models, which can assign paragraphs to the correct categories according to their features.

[0086] S1-5, Recursive Subdivision: For some paragraphs containing multiple sub-clauses, adopt the method of recursive subdivision to further split the paragraphs into smaller units until the predetermined subdivision degree is reached.

[0087] S1-6, Logical Verification: After the splitting is completed, conduct logical verification to ensure the integrity and logical coherence of each paragraph. This step may require manual review to correct errors that may be generated by the algorithm.

[0088] S1-7, Result Output: Finally, output the split paragraphs in the form of structured data, with each paragraph carrying corresponding labels and serial numbers for subsequent analysis and processing.

[0089] Through the above steps, the contract content can be refined into the following twenty main paragraphs:

[0090] Title and Preface: Include basic information such as the contract name, names of both parties, purpose and basis for signing the contract.

[0091] Contract Content: Outline the main content of the contract and clarify the core terms of the transaction between the two parties.

[0092] Service Requirements: Specifically describe the service content, standards, scope, and service quality requirements of the service provider.

[0093] Contract term: Clearly define the valid term of the contract, including the start and end dates.

[0094] Amount and payment method: Detail the total contract price, payment method, payment time nodes and other financial terms.

[0095] Invoice information: Specify the issuance, delivery of invoices and related tax matters.

[0096] Delivery terms: Agree on the time, place, method of delivering goods or providing services and the transfer of liability.

[0097] Acceptance criteria: Establish the acceptance procedures, criteria and the rights and obligations of both parties during the acceptance process.

[0098] Quality assurance: Specify the quality assurance period and remedial measures for the products or services provided by Party B.

[0099] Rights and obligations of Party A: Detail the rights and obligations of Party A during the performance of the contract.

[0100] Rights and obligations of Party B: Detail the rights and obligations of Party B during the performance of the contract.

[0101] Force majeure factors: Define the scope, consequences of force majeure events and the handling methods of both parties.

[0102] Liability for breach of contract: Clearly define the determination of breach of contract, liability for breach of contract and the calculation method of liquidated damages.

[0103] Dispute resolution: Agree on the method, place of resolving contract disputes and the applicable law.

[0104] Quality assessment measures: If any, specify the specific measures and criteria for assessing service quality.

[0105] Contract signing time and place: Record the specific time and place of contract signing.

[0106] Contract effectiveness description: Explain the conditions and procedures for the contract to become effective.

[0107] Alteration and termination: Specify the conditions and procedures for contract alteration and termination.

[0108] Intellectual property rights: Include clauses related to the ownership, use of intellectual property rights and liability for infringement.

[0109] Confidentiality clause: Agree on the scope of protection and confidentiality period for the business secrets of both parties.

[0110] S2. The contract text is converted into structured data units, which is convenient for large models to conduct precise text analysis and extract risk elements, thus laying a solid foundation for subsequent risk audit work.

[0111] S3. Table text conversion: Convert the table data in the contract into a plain text format that can be processed by large models. This step is crucial for ensuring the integrity of the contract content and improving the accuracy of text analysis. As Figure 3 shown, the specific implementation method is as follows:

[0112] S2-1. Table recognition: First, use image recognition technology to locate and recognize the table area in the contract document. Utilize computer vision algorithms, such as convolutional neural networks (CNNs), to analyze the contract scanned image to distinguish text and table content.

[0113] S2-2. Table structure parsing: Parse the recognized table structure, including the division of rows, columns, and cells. This step requires identifying complex structures such as table lines, cell boundaries, and merged cells to accurately extract table data.

[0114] S2-3. Data extraction: Extract text data from the parsed table. This process involves optical character recognition (OCR) technology to convert the image text in the table into an editable text format. For parts that are difficult to recognize by OCR, use deep learning models for character recognition and correction.

[0115] S2-4. Format conversion: Convert the extracted table data into a unified text format, such as JSON or XML, for subsequent text analysis and data processing. During the conversion process, retain the original structure information of the table data, such as column headers and row labels, to reproduce the logical relationship of the table in the text.

[0116] S2-5. Data verification: Verify the converted text data to ensure data accuracy and integrity. This step may include comparing the original table and the converted text to check for data loss or errors.

[0117] S2-6. Text integration: Integrate the table text-converted data with the contract body text to form a complete contract text, providing a unified data source for subsequent text semantic analysis and element verification.

[0118] Table text conversion is an essential part of this method. It realizes the effective conversion of contract table data through advanced technical means, providing comprehensive and accurate data support for contract risk auditing.

[0119] S4. Checklist table construction: Systematically organize the key information in the contract to form a structured checklist table for subsequent risk identification and analysis. As Figure 4 shown, the specific implementation of this step is as follows:

[0120] S4-1. Information extraction: Based on the contract text, extract the core elements in the contract, including but not limited to the contract parties, subject matter, quantity, quality, price, performance period, payment method, liability for breach of contract, etc. These elements are the cornerstones of contract execution and risk management;

[0121] S4-2. Element classification: Classify the extracted information according to the nature and function of the contract terms, such as contract party information, transaction conditions, rights and obligations clauses, liability for breach of contract clauses, etc., to ensure the logic and organization of the list table.

[0122] S4-3. Table design: According to the extracted and classified information, design the framework structure of the list table. The table should contain all necessary columns and rows to accommodate different contract elements and maintain sufficient scalability to adapt to the characteristics of different contracts.

[0123] S4-4. Data filling: Fill the extracted and classified information into the corresponding positions in the list table to ensure that each element is accurately mapped into the table. This step requires a high degree of precision to avoid information misalignment or omission.

[0124] S4-5. Verification and review: Verify the filled list table to ensure that all the entered data is consistent with the original contract text without omission or error. This process may involve manual review to supplement the deficiencies of automated processing.

[0125] S4-6. Format optimization: Optimize the format of the list table to make it both meet the audit requirements and be convenient for reading and analysis. Format optimization includes adjusting column widths, row heights, font sizes, etc. to improve the readability of the table.

[0126] The construction of the list table is a key step in realizing the automation and intelligentization of contract risk audit in this method, and it is of great significance for improving the professional level and efficiency of contract risk management.

[0127] S5. Text semantic analysis: Through in-depth understanding of the contract text, identify the potential risks and meanings in the contract terms; as Figure 5 shown, this step uses advanced natural language processing (NLP) technology, and the specific implementation is as follows:

[0128] S5-1. Word segmentation and part-of-speech tagging: First, perform word segmentation on the contract text to split the continuous text into independent lexical units and tag the part of speech of each word, such as nouns, verbs, adjectives, etc., to lay the foundation for subsequent semantic analysis.

[0129] S5-2. Syntactic analysis: Through syntactic analysis of the contract text, construct the syntactic structure tree of the sentence, identify the syntactic relationships such as subject, predicate, object, attributive, adverbial, complement, etc., so as to deeply understand the meaning of the sentence.

[0130] S5-3, Semantic Role Labeling: Identify semantic roles in sentences, such as Party A, Party B, supervisor, etc. This helps to understand the rights and obligations of each party in the contract terms.

[0131] S5-4, Sentiment Analysis: Conduct sentiment analysis on the contract text to judge the sentiment tendency of the clause expression, such as positive, negative or neutral, providing a basis for risk assessment from the sentiment dimension.

[0132] S5-5, Entity Recognition and Relationship Extraction: Use Named Entity Recognition (NER) technology to identify key entities in the contract, such as organizations, locations, times, etc., and extract the relationships between entities, such as subordination, coordination, causality, etc.

[0133] S5-6, Logical Relationship Analysis: Analyze the logical relationships between contract clauses, such as conditions, assumptions, results, etc., to reveal the logical structure and dependency relationships between clauses.

[0134] S5-7, Risk Point Identification: Based on the above analysis results, identify potential risk points in the contract, such as ambiguous expressions, contradictory clauses, unreasonable liability allocations, etc.

[0135] Text semantic analysis is an important step in realizing the intelligentization of contract risk auditing in this method. It provides strong technical support for risk identification and assessment by deeply understanding the semantic information of the contract text.

[0136] S6. Element Verification Conduct a detailed review and verification of the key elements in the contract text to discover and correct potential errors and risks; as Figure 6 shown, the specific implementation of this step is as follows:

[0137] S6-1, Comparison Analysis: Compare the key elements in the contract text with the data in the list table to check whether they are consistent. During the comparison process, focus on the accuracy, integrity of the elements and their consistency with the contract text.

[0138] S6-2, Rule Verification: Verify the contract elements according to the preset rules and standards. These rules may include industry standards, laws and regulations, internal enterprise regulations, etc. During the verification process, for elements that do not conform to the rules, further investigation and analysis are required.

[0139] S6-3, Model Verification: Use a large model to conduct in-depth learning on the contract elements to identify potential risks. By analyzing a large amount of contract data, the large model can discover abnormal situations in the contract elements and provide assistance for verification.

[0140] S6-4, Expert Review: On the basis of model verification, introduce an expert review link. Experts, based on their professional knowledge and experience, conduct further review of the contract elements to improve the accuracy and comprehensiveness of verification.

[0141] S6-5, Feedback and Correction: Based on the results of comparison analysis, rule verification, model verification, and expert review, feedback on the identified problems and correct the relevant elements in the contract text. During the correction process, ensure the fundamental solution to the problems and avoid the recurrence of similar problems.

[0142] S6-6, Verification Record: Record the problems found during the verification process, the handling measures, and the correction results to form a verification record form. The verification record form helps to track the problem-solving situation and provides reference for subsequent audits.

[0143] Element verification is an important step in realizing the intelligence and refinement of contract risk audit in this method, and is of great significance for improving the professional level and efficiency of contract risk management.

[0144] S7, Risk Information Summary: Summarize and analyze the potential risks in the contract text to provide decision-making support for contract risk management; as Figure 7 shown, the specific implementation of this step is as follows:

[0145] S7-1, Risk Point Summarization: Based on the results of text semantic analysis and element verification, summarize the potential risk points in the contract text. Risk points include ambiguous expressions, contradictory clauses, unreasonable responsibility assignments, etc., which may have a negative impact on the performance of the contract.

[0146] S7-2, Risk Level Assessment: Assess the risk level of the summarized risk points and classify them into different levels according to the severity and scope of influence of the risk points. Risk level assessment helps to determine the priority of risk management and provides a basis for decision-making.

[0147] S7-3, Risk Cause Analysis: Conduct an in-depth analysis of the risk points to find out the causes of the risks. This helps to take targeted measures to reduce the likelihood of risks occurring.

[0148] S7-4, Risk Response Strategy: Propose corresponding risk response strategies for the identified risk points. Risk response strategies may include modifying contract terms, adding safeguard measures, formulating emergency plans, etc.

[0149] S7-5, Risk Information Report: Compile the risk information summarization, risk level assessment, risk cause analysis, and risk response strategies into a complete risk information report. The report format includes forms, charts, etc., for easy reading by auditors and users.

[0150] S7-6, Report Approval and Release: Submit the risk information report to the relevant departments for approval to ensure the accuracy and integrity of the report. After approval, release the risk information report to provide a decision-making basis for contract risk management.

[0151] The summary of risk information is an important step in realizing the intelligentization of contract risk audit and decision-making support in this method, and it is of great significance for improving the professional level and efficiency of contract risk management.

[0152] To sum up, this method integrates advanced technologies such as natural language processing, big data analysis, and deep learning to achieve the automated processing and risk identification of contract texts. Through steps such as splitting contract content paragraphs, textifying tables, constructing list tables, text semantic analysis, element verification, and risk information summary, it provides efficient and accurate technical support for contract risk audit.

[0153] Based on the above method, a contract risk audit device in this embodiment includes: at least one memory and at least one processor;

[0154] The at least one memory is used to store machine-readable programs;

[0155] The at least one processor is used to call the machine-readable program to execute a contract risk audit method.

[0156] The above specific implementation manners are only specific cases of the present invention. The patent protection scope of the present invention includes but is not limited to the above specific implementation manners. Any technical solution that conforms to the above specific implementation manners of the present invention and any appropriate changes or substitutions made by those of ordinary skill in the art shall fall within the patent protection scope of the present invention.

[0157] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirits of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A contract risk audit method, characterized in that, It has the following steps: S1. Splitting of contract content paragraphs, decomposing the contract text into several independent and logically related paragraph units; S2. The contract text is transformed into structured data units for text analysis and risk factor extraction; S3. Table text conversion, converting the table data in the contract into a pure text format that can be processed by large models; S4. Construction of a list table, systematically organizing the key information in the contract to form a structured list table; S5. Text semantic analysis, through in-depth understanding of the contract text, identifying potential risks and meanings in the contract terms; S6. Element verification, carefully examining and verifying the key elements in the contract text to discover and correct potential errors and risks; S7. Summary of risk information, summarizing and analyzing the potential risks in the contract text to provide decision-making support for contract risk management.

2. The contract risk audit method according to claim 1, characterized in that In step S1, it includes: S1-1. Preprocessing, first, preprocess the contract text. In addition, perform sentence splitting on the text, splitting the continuous text into independent sentences; S1-2. Title recognition, using the named entity recognition technology in NLP to recognize the titles and subtitles in the contract; S1-3. Determination of paragraph boundaries, after recognizing the titles, determine the start and end boundaries of the paragraphs according to the positions of the titles; S1-4. Content classification, classify the recognized paragraph content according to the preset paragraph categories; S1-5. Recursive subdivision, for some paragraphs containing multiple sub-clauses, adopt the method of recursive subdivision to further split the paragraphs into smaller units until the predetermined subdivision degree is reached; S1-6. Logical verification, after the splitting is completed, perform logical verification; S1-7. Result output, finally, output the split paragraphs in the form of structured data, and each paragraph is accompanied by corresponding labels and serial numbers.

3. The contract risk auditing method according to claim 1, characterized in that In step S3, it includes: S3-1. Table recognition, first, locate and recognize the table areas in the contract document through image recognition technology, and use computer vision algorithms to analyze the contract scanned images to distinguish text and table content; S3-2. Table structure parsing, parse the structure of the recognized table; S3-3. Data extraction, extract text data from the parsed table, use optical character recognition technology to convert the image text in the table into an editable text format, and for parts that are difficult to recognize by OCR, use a deep learning model for character recognition and correction; S3-4. Format conversion, convert the extracted table data into a unified text format, and retain the original structure information of the table data during the conversion process; S3-5. Data verification, verify the converted text data; S3-6. Text integration, integrate the data after table text conversion with the contract body text to form a complete contract text.

4. A contract risk auditing method according to claim 3, characterized in that, In step S4, it includes: S4-1. Information extraction, extract the core elements in the contract based on the contract text; S4-2. Classify the extracted information according to the nature and function of the contract terms; S4-3. Table design, design the framework structure of the list table according to the extracted and classified information; S4-4, Data filling: Fill the extracted classification information into the corresponding positions in the inventory table; S4-5, Verification and auditing: Verify the filled inventory table to ensure that all the entered data is consistent with the original contract text without omission or error; S4-6, Format optimization: Optimize the format of the inventory table to meet the audit requirements and be convenient for reading and analysis.

5. A contract risk auditing method according to claim 4, characterized in that, In step S5, it includes: S5-1, Word segmentation and part-of-speech tagging: First, perform word segmentation on the contract text to split the continuous text into independent lexical units and tag the part of speech of each word; S5-2, Syntactic analysis: Construct the syntactic structure tree of the sentence by performing syntactic analysis on the contract text; S5-3, Semantic role labeling: Identify the semantic roles in the sentence; S5-4, Sentiment analysis: Perform sentiment analysis on the contract text to judge the sentiment tendency of the clause expression; S5-5, Entity recognition and relationship extraction: Use named entity recognition technology to identify the key entities in the contract and extract the relationships between entities; S5-6, Logical relationship analysis: Analyze the logical relationships between the contract clauses to reveal the logical structure and dependency relationships between the clauses; S5-7, Risk point identification: Based on the above analysis results, identify the potential risk points in the contract.

6. The contract risk auditing method according to claim 5, wherein In step S6, it includes: S6-1, Comparison and analysis: Compare the key elements in the contract text with the data in the inventory table to check whether they are consistent; S6-2, Rule verification: Verify the contract elements according to the preset rules and standards; during the verification process, if elements that do not conform to the rules are found, further investigation and analysis are required; S6-3, Model verification: Use the large model to perform deep learning on the contract elements to identify the potential risks; S6-4, Expert review: On the basis of model verification, introduce the expert review link. Experts further review the contract elements according to their professional knowledge and experience; S6-5, Feedback and correction: According to the results of comparison and analysis, rule verification, model verification, and expert review, feedback the discovered problems and correct the relevant elements in the contract text; S6-6, Verification record: Record the problems, handling measures, and correction results discovered during the verification process to form a verification record form.

7. A contract risk auditing method according to claim 6, characterized in that, In step S7, it includes: S7-1, Risk point summary: Based on the results of text semantic analysis and element verification, summarize the potential risk points in the contract text; S7-2, Risk level assessment: Assess the risk level of the summarized risk points and classify them into different levels according to the severity and impact range of the risk points; S7-3, Risk cause analysis: Conduct in-depth analysis of the risk points to find out the causes of the risks; S7-4, Risk response strategy: Propose corresponding risk response strategies for the identified risk points; S7-5, Risk information report: Compile the risk information summary, risk level assessment, risk cause analysis, and risk response strategy content into a complete risk information report; S7-6, Report approval and release: Submit the risk information report to the relevant department for approval, and after approval, release the risk information report.

8. A contract risk audit device, characterized in that It includes: At least one memory and at least one processor; The at least one memory is used for storing a machine-readable program; The at least one processor is used for calling the machine-readable program to execute the method according to any one of claims 1 to 7.

Images