An Automated Contract Text Review Method and System Based on Unified Clause Modeling

Abstract

This invention discloses an automated contract text review method and system based on unified clause modeling, belonging to the field of contract review technology. The specific steps of the review method are as follows: Ⅰ. Accessing a third-party credit reporting agency or credit rating platform to extract credit information of both parties to the contract for credit investigation and review; Ⅱ. Establishing initial contract templates categorized by economic transactions and extracting key contract information using NLP technology. This invention can quickly identify and select the best rule from a large number of detection rules, significantly improving review efficiency, reducing the time required for manual intervention, making the review process more flexible, adapting to the specific needs of different contracts, and comprehensively identifying potential risks in contracts. It helps reviewers promptly identify high-risk clauses, reduces legal risks in contract performance, protects the interests of universities, and provides more robust risk control and compliance guarantees for universities. Furthermore, through integration with accounting systems, it monitors contract performance, ensures contract compliance and enforceability, and improves the transparency and traceability of contract management.

Description

Technical Field

[0001] This invention relates to the field of contract review technology, and in particular to an automated contract text review method and system based on unified clause modeling. Background Technology

[0002] In the modern business environment, contract management is becoming increasingly complex. In higher education, universities sign contracts with multiple parties in their daily operations, and contract management covers a wide range of areas, including faculty recruitment, research projects, industry-university cooperation, and service procurement. Universities need to handle a large number of contracts, ensuring compliance and mitigating legal risks. The compliance and enforceability of these contracts are crucial to the university's reputation and financial security. However, faced with complex contract terms and diverse business needs, traditional manual review methods are often inadequate for meeting the efficiency requirements of modern university contract management. These methods often rely on manual operation, are inefficient and prone to errors, and result in a cumbersome and time-consuming contract management process. With the increasing complexity of the higher education environment and frequent changes in laws and regulations, there is an urgent need for an efficient and reliable contract review solution.

[0003] A search revealed Chinese patent CN115809653A, which discloses an intelligent contract review method and system. While this invention addresses the challenges of manual contract review in companies, such as high workload and unavoidable errors, and improves review accuracy and efficiency, it does not review contract clauses. This results in higher contract risks, an inability to identify potential legal and financial risks in real time, and an inability to guarantee contract compliance and enforceability. Furthermore, it fails to provide universities with robust risk control and compliance guarantees, and reduces the transparency and traceability of contract management. Therefore, we propose an automated contract text review method and system based on unified clause modeling. Summary of the Invention

[0004] The purpose of this invention is to address the deficiencies in the existing technology by proposing an automated contract text review method and system based on unified clause modeling.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] An automated contract text review method based on unified clause modeling, the specific steps of which are as follows:

[0007] Ⅰ. Access third-party credit reporting agencies or credit rating platforms to extract credit information of both parties to the contract in order to conduct credit investigations and audits of both parties;

[0008] II. Establish initial contract templates categorized by economic transactions and extract key contract information using NLP technology;

[0009] III. Using OCR technology, review key information in the initial contract template to identify risk factors involved in the contract and mark the key points for review in the contract;

[0010] IV. Generate electronic or paper contracts based on the needs of both parties, remind the signatory to confirm all contract terms and risks, and monitor the contract signing status in real time;

[0011] V. Review the completeness of the signed contract information and extract key contract data using NLP and OCR technologies for filing and registration;

[0012] VI. Interact with the financial accounting system to analyze contract performance based on the extracted key contract information, and issue risk alerts based on the performance.

[0013] The automated contract text review method based on unified clause modeling is characterized by the following specific steps in the credit investigation and review described in step I:

[0014] S1.1: Connect with external third-party credit reporting agencies or credit rating platforms, and obtain the credit records, business status, historical litigation information and administrative penalty records of the contracting parties by calling the corresponding API interfaces;

[0015] S1.2: Automatically analyze the financial data of the contracting party through public financial information or financial statements provided by suppliers or customers, automatically retrieve the contracting party's historical contract execution records, and query the contracting party's past contract performance through historical data comparison;

[0016] S1.3: Explore the connections between the contracting party and other enterprises, identify its affiliated companies and upstream and downstream enterprises, and automatically detect the risk information of the contracting party and its affiliates by combining publicly available judicial information and enterprise operation information;

[0017] S1.4: Based on the collected information, the contracting parties are classified into different risk levels according to the internal rules of the risk database. After the credit score and risk rating are generated, a risk assessment report is automatically generated, and potential risk points are highlighted. If the system identifies that the contracting party has a high risk of default, an early warning reminder will be automatically displayed.

[0018] The automated contract text review method based on unified clause modeling is characterized by the following specific steps for extracting key contract information in step II:

[0019] S2.1: Establish standardized contract templates based on different types of economic transactions. In accordance with the latest laws, regulations, industry policies, and the regulations of the universities signing the contracts, the legal department reviews the legality, reasonableness, risk points, and modification suggestions of the clauses in each contract template.

[0020] S2.2: Based on actual business needs, both parties to the contract select the corresponding contract template and adjust and modify the template. The review system extracts the modified contract terms in real time and provides real-time compliance prompts. At the same time, the review system provides modification suggestions and records the change history during the modification process.

[0021] S2.3: Remove redundant spaces, punctuation marks and formatting errors from the contract text, and then use word segmentation technology to split the contract text into individual words or phrases. Extract keywords corresponding to the contract to be tested from a comparison database containing industry keywords and legal terms, and compare each group of keywords with the contract text to identify important content in the text.

[0022] S2.4: Automatically identify important entities in the contract text through named entity recognition technology, and label and store the identified entities. Use syntactic analysis technology to parse the grammatical structure of the contract text, identify the subject, predicate and object components in the sentences, and obtain the logical relationship between the various clauses of the contract.

[0023] S2.5: Extract key clauses from the parsed text and classify them. At the same time, organize the extracted key clauses into a standardized format. Then, use a text similarity algorithm to compare the extracted clauses with historical contract templates to identify similar clauses and differences.

[0024] The automated contract text review method based on unified clause modeling is characterized by the following specific steps in step III for identifying risk factors involved in the contract:

[0025] S3.1: Collect images of the initial contract text drafted by both parties, convert them into a format that the review system can process, extract the text from the image of the contract text after format conversion using OCR technology, and convert the text in the image into an editable text format;

[0026] S3.2: Identify and remove irrelevant characters and noise from the text using regular expressions and text processing functions, then use dictionaries and context information for comparison to identify recognition errors that occur during the OCR process, automatically correct erroneous characters, and then convert all text into a unified character encoding;

[0027] S3.3: Use Chinese word segmentation tools to identify words in the cleaned text and split them into individual words or phrases to form a word list. Identify and remove stop words from the word list. Use a lemmatization algorithm to perform lemmatization on the word segmentation results. Finally, perform text verification through text length check and specific format verification.

[0028] S3.4: Load the standard contract template that has been reviewed by the legal department, compare the extracted contract text with the standard template, analyze the clauses one by one, determine whether there are any missing, changed or inconsistent clauses, and automatically identify the risk factors in the contract based on the preset rules in the risk database.

[0029] S3.5: Combining historical contract data and market data, and utilizing big data analytics, assess the risk factors of the contract, then summarize the identified risk factors and assessment results to generate a detailed risk assessment report.

[0030] The automated contract text review method based on unified clause modeling is characterized by the following specific steps in S3.4 for automatically identifying risk factors in the contract:

[0031] S4.1: The risk library receives the contract text to be reviewed and obtains the current contract review progress. Based on the current review progress, it extracts the required sets of detection rules. Then, it integrates the current review status and review progress into a set of root nodes and uses the new review status generated after applying different detection rules as child nodes. It sets the current risk assessment value, execution cost and expected effect attributes of each node.

[0032] S4.2: Perform a random simulation for each child node to evaluate the different audit results generated after the rule is applied. Through multiple iterations, obtain the success rate and potential risks of rule application and record the results of each simulation. Then, based on the simulation results, select the detection rule with the highest success rate under the current node, expand the tree structure, and generate new child nodes.

[0033] S4.3: During the expansion process, the risk library simultaneously identifies and marks potential risk points. After the expansion is completed, each node in the tree is evaluated, its risk assessment value and execution cost are updated, and nodes that do not meet the preset requirements are pruned according to the evaluation results. Then, based on the tree structure, the path with the highest success rate and the lowest potential risk is selected as the final contract review solution.

[0034] S4.4: After the current audit progress is completed, a detailed risk assessment report is generated, listing the risk factors found in the audit and their corresponding detection rules. The audit results and identified risk information are fed back to the risk database for regular updates and adjustments to the detection rules. At the same time, when the next stage of audit begins, the corresponding audit plan is regenerated based on the next audit status to audit the contract text.

[0035] The automated contract text review system based on unified clause modeling includes a contract investigation module, a risk database management module, a template management module, a clause extraction module, an OCR recognition module, a contract review module, a risk assessment module, a signing and filing module, a performance monitoring module, an analysis report module, and an access control module.

[0036] The contract investigation module is used to conduct credit investigations on both parties to the contract, including assessments of creditworthiness, qualifications, relevance, and ability to perform the contract.

[0037] The risk database management module is used to establish a risk database and set separate detection rules and standards for each review process to gradually identify the risk factors in the contract review process.

[0038] The template management module is used to create and maintain multiple sets of standard contract templates that have passed legal review;

[0039] The clause extraction module is used to extract and structure key clauses from contract texts using NLP technology, forming standardized clause data.

[0040] The OCR recognition module is used to convert paper contracts or image documents into editable text using OCR technology;

[0041] The contract review module is used to conduct a comprehensive review of the extracted contract texts and standard templates, and to identify potentially high-risk clauses.

[0042] The risk assessment module is used to assess risk factors in the contract text based on the contract review results and the detection rules in the risk database, so as to generate a risk report and mark high-risk clauses.

[0043] The signing and filing module is used to record signing information and automatically file contracts.

[0044] The performance monitoring module is used to monitor the performance of the contract in real time based on the extracted key information;

[0045] The analysis report module is used to comprehensively analyze the data of the entire contract review process and generate various management reports;

[0046] The permission management module is used to manage the permissions of system users.

[0047] As a further aspect of the present invention, the specific steps for the risk database management module to establish the risk database are as follows:

[0048] S5.1: Collect historical cases and expert opinions, and identify various risk factors such as legal risks, financial risks and performance risks by analyzing historical contract data and industry practice data. Then, design the risk database to store fields for risk types, risk descriptions, impact levels, detection rules and countermeasures.

[0049] S5.2: For each risk factor, formulate specific detection rules and standards, set trigger conditions and response measures for each rule, and then input the identified risk factors and their corresponding detection rules into the risk database for sorting and classification.

[0050] S5.3: The initial risk library is built by simulating the contract review process. The rules are adjusted and optimized based on the test results. The risk library is then reviewed and updated regularly based on market changes, legal and regulatory updates, university needs and internal feedback. New risk factors and detection rules are added, and content that is no longer applicable is removed. At the same time, risk library usage documents and manuals are generated to explain the functions and usage methods of the risk library in detail.

[0051] S5.4: Automatically integrate the completed risk database with the contract review process, and automatically identify risks based on the review progress and detection rules. When a risk factor is detected, an early warning message is automatically generated to promptly remind contract reviewers to pay attention and handle the situation.

[0052] As a further aspect of the present invention, the signing and filing module automatically performs the following specific steps for contract filing: After the contract is approved, when online signing is selected, an electronic contract containing all relevant terms and information is generated. After the contract is generated, the relevant parties sign it online through the electronic signing platform, while recording the identity information and signing time of each signatory to ensure the legality of the contract. After the electronic signing is completed, the system automatically uploads the contract data to the contract management database and generates a filing record.

[0053] When offline signing is selected, after approval, the contract is printed as a paper copy, ensuring all terms are clearly visible. After printing, the relevant parties sign and stamp the paper contract. Once stamped, the contract is scanned using a scanner or high-quality camera to generate a high-resolution image file. The signing and filing module receives the scanned contract image, removes image noise, and adjusts image contrast and brightness to improve recognition accuracy. Then, OCR technology is used to recognize the contract text, extracting key information including signatory information, contract terms, dates, and other relevant content. After recognition, the system verifies the extracted information, comparing it with standard templates or original contract data to ensure accuracy. The OCR-recognized and verified contract information is stored in the contract management system, generating an electronic filing record. The electronic version of the contract is then linked to the recognized data to ensure the integrity of the contract archive. After filing, the system automatically notifies the relevant parties to confirm the contract filing status and information. Simultaneously, relevant personnel can query the filed contract information through the contract management system.

[0054] As a further aspect of the present invention, the specific steps of the performance monitoring module in real-time monitoring the performance of the contract are as follows:

[0055] S6.1: Extract key information from the signed contract, including payment terms, delivery deadlines, amounts, and bank account information of both parties. Then, use NLP technology to process the contract text to form a standardized data format.

[0056] S6.2: Connect the extracted key information with the financial accounting system through API or database, and update and synchronize the data of each group of contracts in real time. According to the set monitoring frequency, regularly monitor the performance progress of the contract and check the key nodes of each group of payment dates and delivery times stipulated in the contract terms.

[0057] S6.3: Extract actual account information from the financial accounting system and compare it with the contract terms. Automatically detect contract performance based on the risk database detection rules. When a late payment or abnormal payment is detected, automatically generate risk warning information and notify the corresponding contract management personnel through the system. At the same time, regularly generate reports on contract performance and summarize all performance analysis results, risk warnings and handling status.

[0058] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0059] This invention interfaces with external third-party credit reporting agencies or credit rating platforms, and obtains various credit data of the contracting parties by calling corresponding API interfaces to conduct credit investigations on both parties. It then collects historical cases and expert opinions, and establishes a corresponding risk database by analyzing historical contract data and industry practice data. For each risk factor, specific detection rules and standards are formulated for the risk database. The completed risk database is automatically integrated with the contract review process. Based on the review progress and detection rules, corresponding detection schemes are constructed for each review stage to automatically identify risk information at each stage of the contract process. Multiple sets of standard contract templates that have passed legal review are established and maintained. Then, NLP technology and OCR intelligent recognition are used to extract key information from the contract, marking the key review points. When the contract passes review, it is signed online or offline according to the choice of both parties, recording the identity information and signing time of each signatory. Simultaneously, the signed contract data is filed. After filing, the system automatically notifies relevant parties to confirm the filing status and information of the contract. Relevant personnel can also query the filing through the contract management system. The system extracts key information from signed contracts, including payment terms, delivery deadlines, amounts, and bank account information of both parties. Using NLP technology, the contract text is processed to create a standardized data format. This extracted key information is then integrated with the financial accounting system via API or database, ensuring real-time updates and synchronization of contract data. Based on a set monitoring frequency, contract performance progress is monitored regularly, and periodic reports on contract performance are generated. The system summarizes all performance analysis results, risk warnings, and handling statuses. It can quickly identify and select the best rules from a large pool of detection rules, significantly improving review efficiency, reducing the time required for manual intervention, and making the review process more flexible to adapt to the specific needs of different contracts. It can comprehensively identify potential risks in contracts, helping reviewers promptly identify high-risk clauses, reduce legal risks during contract performance, protect the interests of universities, and provide more robust risk control and compliance guarantees for universities. Furthermore, through integration with the accounting system, it monitors contract performance, ensuring contract compliance and enforceability, and improving the transparency and traceability of contract management. Attached Figure Description

[0060] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0061] Figure 1 This is a flowchart of the automated contract text review method based on unified clause modeling proposed in this invention;

[0062] Figure 2 This is a system block diagram of the automated contract text review system based on unified clause modeling proposed in this invention;

[0063] Figure 3 This is a block diagram of the contract management system of the automated contract text review system based on unified clause modeling proposed in this invention. Detailed Implementation

[0064] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0065] Example 1: Refer to Figure 1 , Figure 3 This embodiment discloses an automated contract text review method based on unified clause modeling. The specific steps of this review method are as follows:

[0066] Connect with third-party credit reporting agencies or credit rating platforms to conduct credit investigations and audits of both parties to the contract.

[0067] Specifically, the system connects with external third-party credit reporting agencies or credit rating platforms, and obtains the credit records, business status, historical litigation information, and administrative penalty records of the contracting parties by calling the corresponding API interfaces. It automatically analyzes the financial data of the contracting parties through public financial statements or financial statements provided by suppliers or customers, automatically retrieves the contracting parties' historical contract execution records, and queries the contracting parties' past contract performance through historical data comparison. It also explores the connections between the contracting parties and other enterprises, identifies their affiliated companies and upstream and downstream enterprises, and automatically detects the risk information of the contracting parties and their related parties by combining publicly available judicial information and business information. Based on the collected information, the system classifies the contracting parties into different risk levels according to the internal rules of the risk database. After the credit score and risk rating are generated, a risk assessment report is automatically generated, and potential risk points are highlighted. If the system identifies that the contracting party has a high risk of default, it will automatically pop up an early warning reminder.

[0068] Establish initial contract templates categorized by economic transactions and extract key contract information using NLP technology.

[0069] Specifically, standardized contract templates are established based on different types of economic transactions. In accordance with the latest laws, regulations, and industry policies, the legal department reviews the legality, reasonableness, and risk points of the clauses in each contract template, and provides modification suggestions. Based on actual business needs, both parties select the corresponding contract template and adjust and modify it. The review system extracts modified contract clause information in real time and provides real-time compliance prompts. Simultaneously, the review system provides modification suggestions and records change history during the modification process, removing redundant spaces, punctuation marks, and formatting errors from the contract text. Then, word segmentation technology is used to break the contract text down into individual words or phrases, including those containing industry keywords and legal information. Keywords corresponding to the contract to be tested are extracted from the terminology comparison database. Each set of keywords is compared with the contract text to identify important content in the text. Important entities in the contract text are automatically identified through named entity recognition technology, and the identified entities are labeled and stored. The syntactic structure of the contract text is parsed through syntactic analysis technology to identify the subject, predicate, and object components in the sentences to obtain the logical relationship between the contract clauses. Key clauses are extracted from the parsed text and classified. At the same time, the extracted key clauses are organized into a standardized format. Then, the extracted clauses are compared with historical contract templates using text similarity algorithms to identify similar clauses and differences.

[0070] Using OCR technology, key information in the initial contract template is reviewed to identify risk factors involved in the contract and to highlight key points for review.

[0071] Specifically, the process involves collecting images of the initial contract text drafted by both parties and converting them into a format that the review system can process. OCR technology is then used to extract the text from the converted contract text images, converting the text in the images into editable text. Regular expressions and text processing functions are used to identify and remove irrelevant characters and noise from the text. Dictionary and contextual information are then used for comparison to identify recognition errors that occurred during the OCR process, and erroneous characters are automatically corrected. Finally, all text is converted to a unified character encoding, and a Chinese word segmentation tool is used to identify and break down the words in the cleaned text into individual words or phrases. A word list is generated, stop words are identified and removed, and a lexical reconstruction algorithm is used to reconstruct the word segments. Finally, text verification is performed through text length checks and specific format validation. A standard contract template, reviewed by legal counsel, is loaded, and the extracted contract text is compared with the template. Clauses are analyzed one by one to determine if there are any omissions, changes, or inconsistencies. Based on preset rules in a risk database, risk factors in the contract are automatically identified. Combining historical contract data and market data, big data analytics is used to assess the contract's risk factors. Finally, the identified risk factors and assessment results are summarized to generate a detailed risk assessment report.

[0072] It should be further explained that the risk database receives contract texts awaiting review and obtains the current review progress. Based on the current review progress, it extracts the necessary sets of detection rules. Then, it integrates the current review status and review progress into a root node, and uses the new review status generated after applying different detection rules as child nodes. Each node is assigned a current risk assessment value, execution cost, and expected effect. Random simulations are performed on each child node to evaluate the different review results generated after the rule is applied. Through multiple iterations, the success rate and potential risks of rule application are obtained, and the results of each simulation are recorded. Finally, based on the simulation results, the detection rule with the highest success rate under the current node is selected, the tree structure is expanded, and new child nodes are generated. During the expansion process, the risk database simultaneously identifies and marks potential risk points. After expansion, each node in the tree is evaluated, and its risk assessment value and execution cost are updated. Based on the evaluation results, nodes that do not meet the preset requirements are pruned. Then, based on the tree structure, the path with the highest success rate and lowest potential risk is selected as the final contract review plan. After the current review progress is completed, a detailed risk assessment report is generated, listing the risk factors found in the review and their corresponding detection rules. The review results and identified risk information are fed back to the risk database for regular updates and adjustments to the detection rules. At the same time, when the next stage of review begins, the corresponding review plan is regenerated based on the next review status to review the contract text.

[0073] Generate electronic or paper contracts based on the needs of both parties, remind the signatory to confirm all contract terms and risks, and monitor the contract signing status in real time.

[0074] The integrity of the signed contract information is reviewed, and key contract data is extracted and registered using NLP and OCR technologies.

[0075] It interacts with the financial accounting system to analyze contract performance based on extracted key contract information, and issues risk alerts based on the performance status.

[0076] Example 2: Refer to Figure 2-3 This embodiment discloses an automated contract text review system based on unified clause modeling, including a contract investigation module, a risk database management module, a template management module, a clause extraction module, an OCR recognition module, a contract review module, a risk assessment module, a signing and filing module, a performance monitoring module, an analysis report module, and an access control module.

[0077] The contract investigation module is used to conduct credit investigations on both parties to the contract, including assessments of creditworthiness, qualifications, relevance, and performance capabilities; the risk database management module is used to establish a risk database, and at the same time, to set sub-line detection rules and standards for each review process, gradually identifying the risk factors existing in the contract review process.

[0078] Specifically, historical cases and expert opinions are collected, and historical contract data and industry practice data are analyzed to identify various risk factors, including legal, financial, and performance risks. Then, the risk database is designed with fields for risk types, descriptions, impact levels, detection rules, and countermeasures. For each risk factor, specific detection rules and standards are developed, and trigger conditions and response measures are set for each rule. The identified risk factors and their corresponding detection rules are then input into the risk database, organized, and categorized. The initial risk database is tested through a simulated contract review process, and the rules are adjusted and optimized based on the test results. The risk database is regularly reviewed and updated based on market changes, legal and regulatory updates, university needs, and internal feedback, adding new risk factors and detection rules and removing inapplicable content. Simultaneously, a risk database user manual and documentation are generated, detailing the database's functions and usage. The completed risk database is automatically integrated with the contract review process, automatically identifying risks based on review progress and detection rules. When a risk factor is detected, an automatic warning is generated to promptly alert contract reviewers for attention and action.

[0079] The template management module is used to create and maintain multiple sets of standard contract templates that have passed legal review; the clause extraction module is used to extract and structure key clauses from contract texts using NLP technology to form standardized clause data; the OCR recognition module is used to convert paper contracts or image documents into editable text using OCR technology; the contract review module is used to conduct a comprehensive review of the extracted contract texts and standard templates, and identify potentially high-risk clauses; the risk assessment module is used to assess risk factors in contract texts based on the contract review results and detection rules in the risk database, in order to generate risk reports and mark high-risk clauses.

[0080] The signing and filing module is used to record signing information and automatically file contracts.

[0081] Specifically, after contract approval, when online signing is selected, an electronic contract containing all relevant terms and information is generated. Once generated, the relevant parties sign the contract online through the electronic signing platform, recording the identity information of each signatory and the signing time to ensure the contract's legality. After electronic signing, the system automatically uploads the contract data to the contract management database and generates a filing record. When offline signing is selected, after approval, the contract is printed as a paper copy, ensuring all terms are clearly visible. After printing, the relevant parties sign and seal the paper contract. After sealing, the contract is scanned using a scanner or high-quality camera to generate a high-resolution image file. The signing and filing module receives the scanned contract image. After image processing, noise is removed, and contrast and brightness are adjusted to improve recognition accuracy. Then, OCR technology is used to recognize the contract text and extract key information, including signatory information, contract terms, dates, and other relevant content. After recognition, the system verifies the extracted information by comparing it with standard templates or original contract data to ensure accuracy. The OCR-recognized and verified contract information is then stored in the contract management system, generating an electronic filing record. The electronic version of the contract is then linked to the recognized data to ensure the integrity of the contract archive. Once the filing is complete, the system automatically notifies relevant parties to confirm the filing status and information. At the same time, relevant personnel can query the filed contract information through the contract management system.

[0082] The performance monitoring module is used to monitor the performance of contracts in real time based on the extracted key information.

[0083] Specifically, key information such as payment terms, delivery deadlines, amounts, and bank account information of both parties are extracted from signed contracts. NLP technology is used to process the contract text to form a standardized data format. The extracted key information is then integrated with the financial accounting system via API or database, and the contract data is updated and synchronized in real time. Based on the set monitoring frequency, the contract performance progress is monitored regularly, and key nodes such as payment dates and delivery times stipulated in the contract terms are checked. Actual account information is extracted from the financial accounting system and compared with the contract terms. The contract performance is automatically detected according to the risk database detection rules. When late payments or abnormal payments are detected, risk warning information is automatically generated and the corresponding contract management personnel are notified through the system. At the same time, a contract performance report is generated regularly, summarizing all performance analysis results, risk warnings, and handling status.

[0084] The analysis report module is used to comprehensively analyze the data of the entire contract review process and generate various management reports; the access control module is used to manage the access rights of system users.

Claims

1. An automated contract text review method based on unified clause modeling, characterized in that: The specific steps of this audit method are as follows: Ⅰ. Access third-party credit reporting agencies or credit rating platforms to extract credit information of both parties to the contract in order to conduct credit investigations and audits of both parties; II. Establish initial contract templates categorized by economic transactions and extract key contract information using NLP technology; III. Using OCR technology, review key information in the initial contract template to identify risk factors involved in the contract and mark the key points for review in the contract; IV. Generate electronic or paper contracts based on the needs of both parties, remind the signatory to confirm all contract terms and risks, and monitor the contract signing status in real time; V. Review the completeness of the signed contract information and extract key contract data using NLP and OCR technologies for filing and registration; VI. Interact with the financial accounting system to analyze contract performance based on the extracted key contract information, and issue risk alerts based on the performance. The specific steps for identifying risk factors involved in the contract as described in Step III are as follows: S3.1: Collect images of the initial contract text drafted by both parties, convert them into a format that the review system can process, extract the text from the image of the contract text after format conversion using OCR technology, and convert the text in the image into an editable text format; S3.2: Identify and remove irrelevant characters and noise from the text using regular expressions and text processing functions, then use dictionaries and context information for comparison to identify recognition errors that occur during the OCR process, automatically correct erroneous characters, and then convert all text into a unified character encoding; S3.3: Use Chinese word segmentation tools to identify words in the cleaned text and split them into individual words or phrases to form a word list. Identify and remove stop words from the word list. Use a lemmatization algorithm to perform lemmatization on the word segmentation results. Finally, perform text verification through text length check and specific format verification. S3.4: Load the standard contract template that has been reviewed by the legal department, compare the extracted contract text with the standard template, analyze the clauses one by one, determine whether there are any missing, changed or inconsistent clauses, and automatically identify the risk factors in the contract based on the preset rules in the risk database. S3.5: Combining historical contract data and market data, using big data analytics, assess the risk factors of the contract, and then summarize the identified risk factors and assessment results to generate a detailed risk assessment report; The specific steps for automatically identifying risk factors in a contract as described in S3.4 are as follows: S4.1: The risk library receives the contract text to be reviewed and obtains the current contract review progress. Based on the current review progress, it extracts the required sets of detection rules. Then, it integrates the current review status and review progress into a set of root nodes and uses the new review status generated after applying different detection rules as child nodes. It sets the current risk assessment value, execution cost and expected effect attributes of each node. S4.2: Perform a random simulation for each child node to evaluate the different audit results generated after the rule is applied. Through multiple iterations, obtain the success rate and potential risks of rule application and record the results of each simulation. Then, based on the simulation results, select the detection rule with the highest success rate under the current node, expand the tree structure, and generate new child nodes. S4.3: During the expansion process, the risk library simultaneously identifies and marks potential risk points. After the expansion is completed, each node in the tree is evaluated, its risk assessment value and execution cost are updated, and nodes that do not meet the preset requirements are pruned according to the evaluation results. Then, based on the tree structure, the path with the highest success rate and the lowest potential risk is selected as the final contract review solution. S4.4: After the current audit progress is completed, a detailed risk assessment report is generated, listing the risk factors found in the audit and their corresponding detection rules. The audit results and identified risk information are fed back to the risk database for regular updates and adjustments to the detection rules. At the same time, when the next stage of audit begins, the corresponding audit plan is regenerated based on the next audit status to audit the contract text.

2. The automated contract text review method based on unified clause modeling according to claim 1, characterized in that, The specific steps of the credit investigation and verification described in Step I are as follows: S1.1: Connect with external third-party credit reporting agencies or credit rating platforms, and obtain the credit records, business status, historical litigation information and administrative penalty records of the contracting parties by calling the corresponding API interfaces; S1.2: Automatically analyze the financial data of the contracting party through public financial information or financial statements provided by suppliers or customers, automatically retrieve the contracting party's historical contract execution records, and query the contracting party's past contract performance through historical data comparison; S1.3: Explore the connections between the signatory and other enterprises, identify its affiliated companies and upstream and downstream enterprises, and automatically detect the risk information of the signatory and its related parties by combining publicly available judicial information and enterprise operation information. Then, extract information on the registered capital, shareholders and their investment ratios, legal representatives and key persons in charge of the signatory enterprise, obtain information on the related persons of the enterprise's persons in charge through a third-party database, and compare the extracted information on the enterprise's persons in charge and their related persons with the project leaders of the university. If the names of the persons in charge match, an alert will be issued. S1.4: Based on the collected information, the contracting parties are classified into different risk levels according to the internal rules of the risk database. After the credit score and risk rating are generated, a risk assessment report is automatically generated, and potential risk points are highlighted. If the system identifies that the contracting party has a high risk of default, an early warning reminder will be automatically displayed.

3. The automated contract text review method based on unified clause modeling according to claim 2, characterized in that, The specific steps for extracting key contract information as described in Step II are as follows: S2.1: Establish standardized contract templates based on different types of economic transactions. In accordance with the latest laws, regulations, industry policies, and the regulations of the universities signing the contracts, the legal department reviews the legality, reasonableness, risk points, and modification suggestions of the clauses in each contract template. S2.2: Based on actual business needs, both parties to the contract select the corresponding contract template and adjust and modify the template. The review system extracts the modified contract terms in real time and provides real-time compliance prompts. At the same time, the review system provides modification suggestions and records the change history during the modification process. S2.3: Remove redundant spaces, punctuation marks and formatting errors from the contract text, and then use word segmentation technology to split the contract text into individual words or phrases. Extract keywords corresponding to the contract to be tested from a comparison database containing industry keywords and legal terms, and compare each group of keywords with the contract text to identify important content in the text. S2.4: Automatically identify important entities in the contract text through named entity recognition technology, and label and store the identified entities. Use syntactic analysis technology to parse the grammatical structure of the contract text, identify the subject, predicate and object components in the sentences, and obtain the logical relationship between the various clauses of the contract. S2.5: Extract key clauses from the parsed text and classify them. At the same time, organize the extracted key clauses into a standardized format. Then, use a text similarity algorithm to compare the extracted clauses with historical contract templates to identify similar clauses and differences.

4. An automated contract text review system based on unified clause modeling, used to implement the automated contract text review method based on unified clause modeling as described in any one of claims 1-3, characterized in that, It includes modules for contract investigation, risk database management, template management, clause extraction, OCR recognition, contract review, risk assessment, signing and filing, performance monitoring, analysis report, and access control. The contract investigation module is used to conduct credit investigations on both parties to the contract, including assessments of creditworthiness, qualifications, relevance, and ability to perform the contract. The risk database management module is used to establish a risk database and set separate detection rules and standards for each review process to gradually identify the risk factors in the contract review process. The template management module is used to create and maintain multiple sets of standard contract templates that have passed legal review; The clause extraction module is used to extract and structure key clauses from contract texts using NLP technology, forming standardized clause data. The OCR recognition module is used to convert paper contracts or image documents into editable text using OCR technology; The contract review module is used to conduct a comprehensive review of the extracted contract texts and standard templates, and to identify potentially high-risk clauses. The risk assessment module is used to assess risk factors in the contract text based on the contract review results and the detection rules in the risk database, so as to generate a risk report and mark high-risk clauses. The signing and filing module is used to record signing information and automatically file contracts. The performance monitoring module is used to monitor the performance of the contract in real time based on the extracted key information; The analysis report module is used to comprehensively analyze the data from the entire contract review process and generate various management reports; The permission management module is used to manage the permissions of system users.

5. The automated contract text review system based on unified clause modeling according to claim 4, characterized in that, The specific steps for establishing a risk database in the risk database management module are as follows: S5.1: Collect historical cases and expert opinions, and identify various risk factors such as legal risks, financial risks and performance risks by analyzing historical contract data and industry practice data. Then, design the risk database to store fields for risk types, risk descriptions, impact levels, detection rules and countermeasures. S5.2: For each risk factor, formulate specific detection rules and standards, set trigger conditions and response measures for each rule, and then input the identified risk factors and their corresponding detection rules into the risk database for sorting and classification. S5.3: The initial risk library is built by simulating the contract review process. The rules are adjusted and optimized based on the test results. The risk library is then reviewed and updated regularly based on market changes, legal and regulatory updates, university needs and internal feedback. New risk factors and detection rules are added, and content that is no longer applicable is removed. At the same time, risk library usage documents and manuals are generated to explain the functions and usage methods of the risk library in detail. S5.4: Automatically integrate the completed risk database with the contract review process, and automatically identify risks based on the review progress and detection rules. When a risk factor is detected, an early warning message is automatically generated to promptly remind contract reviewers to pay attention and take action.

6. The automated contract text review system based on unified clause modeling according to claim 4, characterized in that, The specific steps of the performance monitoring module in real-time monitoring the contract performance are as follows: S6.1: Extract key information from the signed contract, including payment terms, delivery deadlines, amounts, and bank account information of both parties. Then, use NLP technology to process the contract text to form a standardized data format. S6.2: Connect the extracted key information with the financial accounting system through API or database, and update and synchronize the data of each group of contracts in real time. According to the set monitoring frequency, regularly monitor the performance progress of the contract and check the key nodes of each group of payment dates and delivery times stipulated in the contract terms. S6.3: Extract actual account information from the financial accounting system and compare it with the contract terms. Automatically detect contract performance based on the risk database detection rules. When a late payment or abnormal payment is detected, automatically generate risk warning information and notify the corresponding contract management personnel through the system. At the same time, regularly generate reports on contract performance and summarize all performance analysis results, risk warnings and handling status.

Images