Electronic contract auditing method and device, computer equipment and storage medium
Through fine-tuning learning of large language models and keyword extraction, combined with electronic contract review prompts, the accuracy of the intelligent contract review system in the fields of law and contract is solved, the automation and intelligence of contract review is realized, and the efficiency and accuracy.
Patent Information
- Application Number
- CN202411253105.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-09
- Publication Date
- 2024-12-06
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing smart contract review system has problems of poor accuracy and poor expansion when handling and reviewing contracts, especially the failure to effectively utilize knowledge in the legal field and contract field, resulting in inaccurate contract review results.
By obtaining historical standard contract data, fine-tuning the pre-trained large language model, extracting keywords and generating electronic contract review prompts, analyzing contract risk situations, forming a standardized audit process and risk assessment system, reducing manual dependence, and improving audit Accuracy and efficiency.
It significantly improves the accuracy and efficiency of contract audits, reduces legal risks and economic losses, enhances the risk management capabilities of enterprises, and realizes the automation and intelligence of contract audits.
Smart Images

Figure CN119090681A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of language processing technology, and in particular to an electronic contract review method, apparatus, computer equipment, computer-readable storage medium, and computer program product. Background Art
[0002] With the continuous development of information technology, contract review has become an indispensable part of commercial contracts. However, the traditional contract review process is usually time-consuming and prone to omissions, and there is a risk of human error. At present, smart contract review systems have been widely used at home and abroad, but these systems generally have some defects and need to be continuously improved and perfected. At the same time, the protection of sensitive information and data is also an important issue.
[0003] At present, there are already some intelligent contract review systems or automated contract review systems based on large language models. They are based on natural language processing technology, but are usually based on vertical domain models or manually designed rules. They can handle limited tasks and fields and have poor scalability. In addition, the contract review is based on the output of the original general knowledge of the large language model, and is not customized based on the knowledge in the legal field and the contract field, resulting in inaccurate contract review results.
[0004] Therefore, there is an urgent need for an electronic contract review method, device, computer equipment, computer-readable storage medium and computer program product that can review electronic contracts based on knowledge in the legal and contract fields and improve the accuracy of contract review results. Summary of the invention
[0005] Based on this, it is necessary to provide an electronic contract review method, device, computer equipment, computer-readable storage medium and computer program product that can review electronic contracts based on knowledge in the legal and contract fields and improve the accuracy of contract review results in response to the above-mentioned technical problems.
[0006] In a first aspect, the present application provides an electronic contract review method, comprising:
[0007] Obtain historical standard contract data;
[0008] Fine-tune the pre-trained large language model based on the historical standard contract data;
[0009] Extract keywords from the electronic contract text using a large language model that has been fine-tuned and learned;
[0010] Get electronic contract review prompts;
[0011] The electronic contract review prompt is used to analyze the keywords to obtain the contract risk situation of the electronic contract text.
[0012] In one embodiment, the obtaining of historical standard contract data includes:
[0013] Access historical contract data in legal verticals;
[0014] Remove sensitive information from historical contract data in the legal vertical field, standardize the contract format, perform multiple rounds of dialogue format conversion, and obtain historical standard contract data.
[0015] In one embodiment, obtaining the electronic contract review prompt includes:
[0016] Obtain audit contract data within the legal vertical;
[0017] Extracting information from the audit contract data to obtain an electronic contract audit prompt;
[0018] Obtaining actual review feedback data after the electronic contract review prompt is applied to the electronic contract text;
[0019] The electronic contract review prompt is optimized based on the actual review feedback data.
[0020] In one embodiment, fine-tuning the pre-trained large language model based on the historical standard contract data includes:
[0021] Based on the historical standard contract data, fine-tune the pre-trained large language model through supervised learning to obtain a model for processing contract terms;
[0022] Transfer learning is performed on the model for processing contract terms to obtain a model for contract review, and the model for contract review is retrained based on the historical standard contract data.
[0023] In one embodiment, obtaining actual review feedback data after the electronic contract review prompt is applied to the electronic contract text includes:
[0024] By simulating an actual contract review scenario, the electronic contract text is reviewed and tested according to the electronic contract review prompts to obtain actual review feedback data of the electronic contract.
[0025] In one embodiment, the audited contract data in the legal vertical field includes public legal documents, legal interpretations, contract samples, historical contract data within the enterprise, and other authorized contract texts.
[0026] In a second aspect, the present application also provides an electronic contract review device, including:
[0027] Acquisition module, used to obtain historical standard contract data;
[0028] A model training module, used for fine-tuning the pre-trained large language model according to the historical standard contract data;
[0029] The keyword extraction module uses a large language model that has been fine-tuned to extract keywords from the electronic contract text;
[0030] A prompt generation module is used to obtain electronic contract review prompts;
[0031] The review module is used to analyze the keywords using the electronic contract review prompt to obtain the contract risk status of the electronic contract text.
[0032] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:
[0033] Obtain historical standard contract data;
[0034] Fine-tune the pre-trained large language model based on the historical standard contract data;
[0035] Extract keywords from the electronic contract text using a large language model that has been fine-tuned and learned;
[0036] Get electronic contract review prompts;
[0037] The electronic contract review prompt is used to analyze the keywords to obtain the contract risk situation of the electronic contract text.
[0038] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:
[0039] Obtain historical standard contract data;
[0040] Fine-tune the pre-trained large language model based on the historical standard contract data;
[0041] Extract keywords from the electronic contract text using a large language model that has been fine-tuned and learned;
[0042] Get electronic contract review prompts;
[0043] The electronic contract review prompt is used to analyze the keywords to obtain the contract risk situation of the electronic contract text.
[0044] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:
[0045] Obtain historical standard contract data;
[0046] Fine-tune the pre-trained large language model based on the historical standard contract data;
[0047] Extract keywords from the electronic contract text using a large language model that has been fine-tuned and learned;
[0048] Get electronic contract review prompts;
[0049] The electronic contract review prompt is used to analyze the keywords to obtain the contract risk situation of the electronic contract text.
[0050] The above-mentioned electronic contract review method, device, computer equipment, computer-readable storage medium and computer program product, by using a pre-trained large language model and fine-tuning it for historical standard contract data, the system can quickly extract keywords from a large number of electronic contract texts, significantly reducing the time for manual reading and analysis of contracts. Fine-tuning learning enables the large language model to better understand the specific context and industry terms of the contract text, thereby improving the accuracy of keyword extraction and the accuracy of subsequent risk analysis. By analyzing the extracted keywords and the electronic contract review prompts, the system can automatically assess the contract risk situation and reduce the reliance on professional legal personnel for risk assessment. Through automated risk assessment, the system can help enterprises promptly discover possible problems and risk points in the contract, so as to take corresponding preventive or remedial measures and enhance the risk management capabilities of the enterprise. The system can form a set of standardized review processes and risk assessment systems based on historical standard contract data to improve the overall standardization level of contract management. As the system learns and analyzes more contract texts, its review capabilities and the accuracy of risk assessment will continue to improve, supporting enterprises to continuously improve the contract review process. The system can help enterprises ensure that the content of the contract complies with the requirements of relevant laws and regulations, and reduce legal risks and economic losses caused by contract violations. In summary, this method realizes the automation and intelligence of the contract review process by combining a large language model and electronic contract review prompts, improves efficiency and accuracy, and reduces costs and risks. BRIEF DESCRIPTION OF THE DRAWINGS
[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.
[0052] FIG1 is an application environment diagram of an electronic contract review method in one embodiment;
[0053] FIG2 is a schematic diagram of a flow chart of an electronic contract review method in one embodiment;
[0054] FIG3 is a schematic diagram of a flow chart of an electronic contract review method in another embodiment;
[0055] FIG4 is a structural block diagram of an electronic contract review device in one embodiment;
[0056] FIG. 5 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION
[0057] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0058] The electronic contract review method provided in the embodiment of the present application can be applied in the application environment shown in FIG1 . In which, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers.
[0059] Server 104 obtains historical standard contract data; fine-tunes the pre-trained large language model based on the historical standard contract data; uses the fine-tuned large language model to extract keywords from the electronic contract text; obtains electronic contract review prompts; uses the electronic contract review prompts to analyze keywords and obtain the contract risk status of the electronic contract text.
[0060] The terminal 102 may be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, IoT devices, and portable wearable devices. The IoT devices may be smart speakers, smart TVs, smart air conditioners, smart car devices, projection devices, etc. Portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. Head-mounted devices may be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc. The server 104 may be an independent physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server that provides cloud computing services.
[0061] In an exemplary embodiment, as shown in FIG2 , an electronic contract review method is provided, which is described by taking the method applied to the server in FIG1 as an example, and includes the following steps S202 to S210. Among them:
[0062] Step S202, obtaining historical standard contract data.
[0063] Specifically, historical standard contract data is obtained to train and fine-tune the large language model so that it can understand and process contract terms. This data will help the system learn the format, terms, legal terms, etc. of the contract, thereby improving the accuracy and efficiency of smart contract review. Data sources include public legal documents, legal interpretations, contract samples, historical contract data within the enterprise, and contract texts authorized from partners.
[0064] Step S204: fine-tune the pre-trained large language model based on historical standard contract data.
[0065] Specifically, pre-trained large language models have been trained on large-scale text datasets and have learned common features of language, including vocabulary, grammar, syntactic structure, etc. These models are able to understand common patterns and contextual relationships of language.
[0066] The purpose of fine-tuning is to adapt the pre-trained model to a specific application scenario, such as contract review. By further training on data in a specific field, the model can learn the professional terminology, expression habits, and text structure of that field. Historical standard contract data provides a wealth of examples that are directly related to contract review. These data contain common clauses, legal terms, format specifications, etc. in contracts, and are key resources for training models to understand contract texts.
[0067] Collect historical standard contract data that covers a variety of contract types and industries to ensure that the model is widely applicable to different contract review scenarios. Fine-tune the pre-trained model using the processed historical standard contract data. During this process, the model's weights and parameters are adjusted based on the contract data to better capture the characteristics and patterns of the contract text.
[0068] Step S206, extracting keywords from the electronic contract text using the large language model that has been fine-tuned and learned.
[0069] Specifically, the large language model after fine-tuning has acquired a deep understanding of texts in specific fields (such as legal contracts). This means that the model not only understands the general features of language, but also learns contract-specific terms, expressions, and structures.
[0070] Keyword extraction is an important step in information retrieval, text analysis, and content understanding. In the context of contract review, keyword extraction can help quickly identify the core content and key terms of the contract, such as the parties to the contract, rights and obligations, payment terms, and liability for breach of contract.
[0071] First, the electronic contract text is input into the fine-tuned large language model. The model analyzes the text content and identifies representative and important words and phrases. This may involve natural language processing technologies such as part-of-speech tagging, entity recognition, and semantic analysis. The model extracts keywords from the text according to preset rules or algorithms. These rules or algorithms may be based on factors such as word frequency, location, and contextual relevance. The model outputs a list of extracted keywords, which can be used for subsequent text analysis, contract review, or information retrieval.
[0072] The extracted keywords can help users quickly understand the main content and key points of the contract and improve the efficiency of contract review. Keyword extraction can also serve as the basis for contract content analysis, risk assessment and compliance inspection, providing support for contract management.
[0073] Step S208, obtaining the electronic contract review prompt.
[0074] Specifically, first, the text data of the electronic contract needs to be input into the system. This data may come from PDF, Word documents or other electronic formats. The input text is preprocessed, including removing irrelevant information, standardizing formats, and word segmentation, etc., to facilitate subsequent analysis. Use NLP technology to automatically identify key terms in the contract, such as contract term, payment terms, liability for breach of contract, confidentiality agreement, etc. Compare the identified key terms with relevant laws and regulations, industry standards or company policies to check whether the contract meets compliance requirements. Analyze the contract terms and assess potential legal risks, such as unfair terms, ambiguous expressions, and excessive liquidated damages. Generate specific audit prompts based on the results of compliance checks and risk assessments. These prompts may include suggested modified terms, risk points that need attention, recommended improvement measures, etc. The generated prompts are presented to users (such as legal advisors or contract managers), who can conduct further review and modification based on these prompts.
[0075] Step S210, using the electronic contract review prompt to analyze keywords, to obtain the contract risk status of the electronic contract text.
[0076] Specifically, the electronic contract text is first cleaned and preprocessed, including removing irrelevant information and standardizing the text format, so as to facilitate subsequent analysis. Natural language processing (NLP) technology is used to automatically identify key information in the contract, such as the parties to the contract, contract term, payment terms, liability for breach of contract, etc. The extracted key information is compared with relevant laws, regulations, industry standards or company policies to check whether the contract meets compliance requirements. Based on the extracted information and the results of the compliance check, the risks in the contract are assessed, such as unfair terms, ambiguous expressions, excessive liquidated damages, etc.
[0077] In the above-mentioned electronic contract review method, by using a pre-trained large language model and fine-tuning it for historical standard contract data, the system can quickly extract keywords from a large number of electronic contract texts, significantly reducing the time for manual reading and analysis of contracts. Fine-tuning learning enables the large language model to better understand the specific context and industry terms of the contract text, thereby improving the accuracy of keyword extraction and the accuracy of subsequent risk analysis. By analyzing the extracted keywords and the electronic contract review prompts, the system can automatically assess the contract risk situation and reduce the reliance on professional legal personnel for risk assessment. Through automated risk assessment, the system can help enterprises promptly discover possible problems and risk points in the contract, so as to take corresponding preventive or remedial measures and enhance the risk management capabilities of the enterprise. The system can form a standardized review process and risk assessment system based on historical standard contract data to improve the overall standardization level of contract management. As the system learns and analyzes more contract texts, its review capabilities and the accuracy of risk assessment will continue to improve, supporting enterprises to continuously improve the contract review process. The system can help enterprises ensure that the content of the contract complies with the requirements of relevant laws and regulations and reduce legal risks and economic losses caused by contract violations. In summary, this method realizes the automation and intelligence of the contract review process by combining a large language model and electronic contract review prompts, improves efficiency and accuracy, and reduces costs and risks.
[0078] In an exemplary embodiment, obtaining historical standard contract data includes:
[0079] Access historical contract data in legal verticals;
[0080] Remove sensitive information from historical contract data in the legal vertical field, standardize the contract format, perform multiple rounds of dialogue format conversion, and obtain historical standard contract data.
[0081] Specifically, obtaining historical contract data in the legal vertical refers to collecting past contract samples from legal-related fields. These contracts may come from court cases, public business transaction records, law firm archives, etc. The collected data should be representative and cover different types of contracts, such as lease contracts, labor contracts, sales contracts, etc.
[0082] Before using contract data, you need to ensure that all personal or sensitive information is removed or anonymized to comply with data protection regulations and privacy rights. Sensitive information may include personal name, address, ID number, bank account information, etc.
[0083] Since contracts may exist in different formats (such as PDF, Word documents, scanned images, etc.), these different formats of contracts need to be converted into a unified format for subsequent processing. Standardization also includes ensuring the structure of the contract content, such as arranging the various terms and conditions of the contract in a certain order and format.
[0084] Multi-round dialogue format conversion refers to converting the contract text into a format suitable for machine learning and natural language processing. For example, the contract terms can be converted into question-answer pairs using a dialogue system. This format conversion helps develop artificial intelligence systems that can understand and process contract texts, enabling them to explain the contract content or provide consulting services by simulating dialogue.
[0085] After the above steps, a set of historical standard contract data that has been cleaned, standardized and formatted is obtained.
[0086] In this embodiment, by removing sensitive information, personal privacy and business secrets are ensured not to be disclosed, which complies with the requirements of data protection regulations. Standardized contract formats make contract data of different sources and types consistent, facilitating unified processing and analysis. Formatted and structured data can be more easily read and processed by computer systems, improving the efficiency and accuracy of data processing. Converting contract data into a multi-round dialogue format enables natural language processing (NLP) technology to be better applied to the analysis and understanding of contract texts.
[0087] In an exemplary embodiment, as shown in FIG3 , obtaining an electronic contract review prompt includes:
[0088] Step S302, obtaining audit contract data in the legal vertical field;
[0089] Step S304, extracting information from the audit contract data to obtain an electronic contract audit prompt;
[0090] Step S306, obtaining actual review feedback data after the electronic contract review prompt is applied to the electronic contract text;
[0091] Step S308, optimizing the electronic contract review prompts based on actual review feedback data.
[0092] Specifically, obtain audit contract data in the legal vertical field: First, it is necessary to collect contract samples from legal-related fields. These data may come from public legal documents, historical contract cases, law firm archives, etc.
[0093] Extract information from audited contract data: Use natural language processing (NLP) technology to automatically identify and extract key information in the contract, such as the parties to the contract, contract term, payment terms, liability for breach of contract, etc.
[0094] Obtain electronic contract audit prompts: Based on the extracted information, generate specific audit prompts that can point out possible problems or risks in the contract and provide intelligent assessment and suggestions for the compliance and security of the contract.
[0095] Actual review feedback data after application to electronic contract texts: Apply these prompts to actual electronic contract texts and collect feedback data from users or legal staff.
[0096] Optimize electronic contract review prompts based on actual review feedback data: Based on the feedback collected, continuously optimize and adjust the review prompts to improve their accuracy and practicality.
[0097] In this embodiment, the workload of manual review is reduced and the review efficiency is improved through automated information extraction and review prompt generation. Automated review prompts can help legal personnel find issues that may be omitted or overlooked in the contract, thereby improving the quality of contract review. The system can identify potential risk points in the contract, provide early warnings, and help companies take preventive measures to reduce legal risks. By collecting actual review feedback data, the system can continuously learn and optimize, making the review prompts more accurate and practical.
[0098] In an exemplary embodiment, fine-tuning the pre-trained large language model based on historical standard contract data includes:
[0099] Based on historical standard contract data, the pre-trained large language model is fine-tuned through supervised learning to obtain a model for processing contract terms.
[0100] Transfer learning is performed on the model used to process contract terms to obtain a model for contract review, and the model for contract review is retrained based on historical standard contract data.
[0101] Specifically, a pre-trained large language model means that it is trained using contract text that has been annotated by legal experts, and it is able to identify key terms and conditions in the contract. The pre-trained large language model will be fine-tuned through supervised learning. Supervised learning is a machine learning method in which the large language model learns from historical standard contract data. During the fine-tuning process, the model adjusts its internal parameters to better understand and process the contract text.
[0102] Once the model has been fine-tuned with supervised learning, it can be used for transfer learning. Transfer learning is a technique that allows a model that has been trained on one task to be applied to a related but different task. In this scenario, the model will be transferred from the task of processing contract terms to the task of contract review. This means that the model will learn how to identify potential problems, inconsistencies, or risk points in the contract.
[0103] After transfer learning, the model will be retrained using more historical standard contract data. The purpose of this training is to further improve the model's performance on contract review tasks. This may include identifying specific risk points in the contract, ensuring compliance with contract terms, checking the integrity of the contract, etc.
[0104] Throughout the training process, the model is continuously optimized to improve its accuracy and efficiency in actual contract review tasks. This involves adjusting the model's hyperparameters, improving the training algorithm, or adding more training data.
[0105] In this embodiment, in this way, the automation level of contract review can be improved, the workload of manual review can be reduced, and the quality and consistency of review can be improved. In addition, this method can also help legal professionals discover issues that may be overlooked, thereby better managing and reducing contract risks.
[0106] In an exemplary embodiment, obtaining actual review feedback data after the electronic contract review prompt is applied to the electronic contract text includes:
[0107] By simulating actual contract review scenarios, the electronic contract text is reviewed and tested according to the electronic contract review prompts to obtain actual review feedback data for the electronic contract.
[0108] Specifically, simulating actual contract review scenarios means creating an environment that mimics as closely as possible the real-world situations in which legal professionals review electronic contracts. This includes using real contract samples, simulating various situations that may be encountered during the review process, and considering common concerns in contract review.
[0109] Applying electronic contract review prompts means using pre-generated electronic contract review prompts in this simulation environment to guide the contract review process. These prompts may include review suggestions for specific clauses, identification of risk points, and legal compliance checks.
[0110] Through simulated review, the electronic contract text is carefully checked using audit prompts. This may involve evaluating the completeness, accuracy and legality of the contract content against legal clauses, contract standards and best practices.
[0111] During the mock review, record the reviewer's response to the audit prompts, feedback, and suggestions for improvement. This data may include comments on the accuracy of the prompts, feedback on the usefulness of the prompts in the actual review, and any situations that may cause review delays or errors.
[0112] The actual review feedback data collected will be used to analyze the performance of the review prompts and identify any areas for improvement. This may involve adjusting the content of the prompts, improving the presentation of the prompts, or enhancing the algorithm used by the model to generate the prompts.
[0113] In this embodiment, the effectiveness is verified by testing actual application scenarios. This simulation test method can help developers and legal professionals ensure the practicality and accuracy of the audit prompts in actual work, thereby improving the quality and efficiency of electronic contract review. In this way, the audit prompts can be continuously optimized to better serve the needs of contract review.
[0114] In an exemplary embodiment, the audited contract data in the legal vertical field includes public legal documents, legal interpretations, contract samples, historical contract data within the enterprise, and other authorized contract texts.
[0115] Specifically, public legal documents include laws and regulations, judicial interpretations, policy guidance documents, etc., which provide a legal basis and standards for contract review. Legal interpretation refers to the interpretation of legal provisions by legal experts, courts or official agencies. These interpretations help understand the specific application of the law and the legality of contract terms. Contract samples refer to standardized or templated contract samples. They show the common terms and structures of different types of contracts and can be used as a reference during the review. The internal historical contract data of an enterprise refers to the contract texts signed by the enterprise in the past. These data reflect the actual business operations and contract habits of the enterprise and are very important for understanding the contract characteristics of a specific industry or field. Other authorized contract texts include contract texts obtained from partners, industry associations or other credible sources. These texts need to be used under the premise of legality and authorization.
[0116] In this embodiment, by integrating these data sources, a comprehensive and powerful contract review model can be constructed, thereby improving the efficiency and accuracy of contract review.
[0117] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.
[0118] Based on the same inventive concept, the embodiment of the present application also provides an electronic contract review device for implementing the electronic contract review method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more electronic contract review device embodiments provided below can refer to the limitations of the electronic contract review method above, and will not be repeated here.
[0119] In an exemplary embodiment, as shown in FIG4 , an electronic contract review device is provided, including: an acquisition module 402 for acquiring historical standard contract data;
[0120] A model training module 404 is used to fine-tune the pre-trained large language model based on historical standard contract data;
[0121] The keyword extraction module 406 extracts keywords from the electronic contract text using the large language model after fine-tuning and learning;
[0122] A prompt generation module 408 is used to obtain an electronic contract review prompt;
[0123] The review module 410 is used to analyze keywords using the electronic contract review prompts to obtain the contract risk status of the electronic contract text.
[0124] In an exemplary embodiment, the acquisition module 402 is also used to obtain historical contract data in the legal vertical field; remove sensitive information in the historical contract data in the legal vertical field, standardize the contract format, perform multiple rounds of dialogue format conversion, and obtain historical standard contract data.
[0125] In an exemplary embodiment, the prompt generation module 408 is also used to obtain audit contract data in the legal vertical field; extract information from the audit contract data to obtain electronic contract audit prompts; obtain actual review feedback data after the electronic contract audit prompts are applied to the electronic contract text; and optimize the electronic contract audit prompts based on the actual review feedback data.
[0126] In an exemplary embodiment, the model training module 404 is used to fine-tune the pre-trained large language model through supervised learning based on historical standard contract data to obtain a model for processing contract terms; transfer learning is performed on the model for processing contract terms to obtain a model for contract review, and the model for contract review is re-trained based on historical standard contract data.
[0127] In an exemplary embodiment, the prompt generation module 408 is also used to simulate an actual contract review scenario, conduct an audit test on the electronic contract text according to the electronic contract review prompts, and obtain actual review feedback data of the electronic contract.
[0128] In an exemplary embodiment, the audited contract data in the legal vertical field includes public legal documents, legal interpretations, contract samples, historical contract data within the enterprise, and other authorized contract texts.
[0129] Each module in the electronic contract review device can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each module.
[0130] In an exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be shown in FIG5. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected via a system bus, and the communication interface is connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store historical standard contract data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, an electronic contract review method is implemented.
[0131] Those skilled in the art will understand that the structure shown in FIG. 5 is merely a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.
[0132] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:
[0133] Obtain historical standard contract data;
[0134] Fine-tune the pre-trained large language model based on historical standard contract data;
[0135] Extract keywords from the electronic contract text using a large language model that has been fine-tuned and learned;
[0136] Get electronic contract review prompts;
[0137] Use electronic contract review prompts to analyze keywords and obtain the contract risk status of the electronic contract text.
[0138] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:
[0139] Access historical contract data in legal verticals;
[0140] Remove sensitive information from historical contract data in the legal vertical field, standardize the contract format, perform multiple rounds of dialogue format conversion, and obtain historical standard contract data.
[0141] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:
[0142] Obtain audit contract data within the legal vertical;
[0143] Extract information from audit contract data and obtain electronic contract audit prompts;
[0144] Obtaining actual review feedback data after the electronic contract review prompt is applied to the electronic contract text;
[0145] Based on actual review feedback data, the electronic contract review prompts are optimized.
[0146] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:
[0147] Based on historical standard contract data, the pre-trained large language model is fine-tuned through supervised learning to obtain a model for processing contract terms.
[0148] Transfer learning is performed on the model used to process contract terms to obtain a model for contract review, and the model for contract review is retrained based on historical standard contract data.
[0149] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:
[0150] By simulating actual contract review scenarios, the electronic contract text is reviewed and tested according to the electronic contract review prompts to obtain actual review feedback data for the electronic contract.
[0151] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:
[0152] Audited contract data in the legal vertical field include public legal documents, legal interpretations, contract samples, historical contract data within the enterprise, and other authorized contract texts.
[0153] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:
[0154] Obtain historical standard contract data;
[0155] Fine-tune the pre-trained large language model based on historical standard contract data;
[0156] Extract keywords from the electronic contract text using a large language model that has been fine-tuned and learned;
[0157] Get electronic contract review prompts;
[0158] Use electronic contract review prompts to analyze keywords and obtain the contract risk status of the electronic contract text.
[0159] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:
[0160] Access historical contract data in legal verticals;
[0161] Remove sensitive information from historical contract data in the legal vertical field, standardize the contract format, perform multiple rounds of dialogue format conversion, and obtain historical standard contract data.
[0162] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:
[0163] Obtain audit contract data within the legal vertical;
[0164] Extract information from audit contract data and obtain electronic contract audit prompts;
[0165] Obtaining actual review feedback data after the electronic contract review prompt is applied to the electronic contract text;
[0166] Based on actual review feedback data, the electronic contract review prompts are optimized.
[0167] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:
[0168] Based on historical standard contract data, the pre-trained large language model is fine-tuned through supervised learning to obtain a model for processing contract terms.
[0169] Transfer learning is performed on the model used to process contract terms to obtain a model for contract review, and the model for contract review is retrained based on historical standard contract data.
[0170] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:
[0171] By simulating actual contract review scenarios, the electronic contract text is reviewed and tested according to the electronic contract review prompts to obtain actual review feedback data for the electronic contract.
[0172] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:
[0173] Audited contract data in the legal vertical field include public legal documents, legal interpretations, contract samples, historical contract data within the enterprise, and other authorized contract texts.
[0174] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:
[0175] Obtain historical standard contract data;
[0176] Fine-tune the pre-trained large language model based on historical standard contract data;
[0177] Extract keywords from the electronic contract text using a large language model that has been fine-tuned and learned;
[0178] Get electronic contract review prompts;
[0179] Use electronic contract review prompts to analyze keywords and obtain the contract risk status of the electronic contract text.
[0180] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:
[0181] Access historical contract data in legal verticals;
[0182] Remove sensitive information from historical contract data in the legal vertical field, standardize the contract format, perform multiple rounds of dialogue format conversion, and obtain historical standard contract data.
[0183] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:
[0184] Obtain audit contract data within the legal vertical;
[0185] Extract information from audit contract data and obtain electronic contract audit prompts;
[0186] Obtaining actual review feedback data after the electronic contract review prompt is applied to the electronic contract text;
[0187] Based on actual review feedback data, the electronic contract review prompts are optimized.
[0188] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:
[0189] Based on historical standard contract data, the pre-trained large language model is fine-tuned through supervised learning to obtain a model for processing contract terms.
[0190] Transfer learning is performed on the model used to process contract terms to obtain a model for contract review, and the model for contract review is retrained based on historical standard contract data.
[0191] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:
[0192] By simulating actual contract review scenarios, the electronic contract text is reviewed and tested according to the electronic contract review prompts to obtain actual review feedback data for the electronic contract.
[0193] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:
[0194] Audited contract data in the legal vertical field include public legal documents, legal interpretations, contract samples, historical contract data within the enterprise, and other authorized contract texts.
[0195] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.
[0196] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.
[0197] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.
[0198] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.
Claims
1. An electronic contract review method, characterized in that: The method comprises: Obtain historical standard contract data; Fine-tune the pre-trained large language model based on the historical standard contract data; Extract keywords from the electronic contract text using a large language model that has been fine-tuned and learned; Get electronic contract review prompts; The electronic contract review prompt is used to analyze the keywords to obtain the contract risk situation of the electronic contract text.
2. The method according to claim 1, characterized in that The obtaining of historical standard contract data includes: Access historical contract data in legal verticals; Remove sensitive information from historical contract data in the legal vertical field, standardize the contract format, perform multiple rounds of dialogue format conversion, and obtain historical standard contract data.
3. The method according to claim 1, characterized in that: The obtaining of electronic contract review prompts includes: Obtain audit contract data within the legal vertical; Extracting information from the audit contract data to obtain an electronic contract audit prompt; Obtaining actual review feedback data after the electronic contract review prompt is applied to the electronic contract text; The electronic contract review prompt is optimized based on the actual review feedback data.
4. The method according to claim 1, characterized in that: The fine-tuning and learning of the pre-trained large language model according to the historical standard contract data includes: Based on the historical standard contract data, fine-tune the pre-trained large language model through supervised learning to obtain a model for processing contract terms; Transfer learning is performed on the model for processing contract terms to obtain a model for contract review, and the model for contract review is retrained based on the historical standard contract data.
5. The method according to claim 3, characterized in that: The obtaining of actual review feedback data after the electronic contract review prompt is applied to the electronic contract text includes: By simulating an actual contract review scenario, the electronic contract text is reviewed and tested according to the electronic contract review prompts to obtain actual review feedback data of the electronic contract.
6. The method according to claim 2, characterized in that The audited contract data in the legal vertical field include public legal documents, legal interpretations, contract samples, historical contract data within the enterprise, and other authorized contract texts.
7. An electronic contract review device, characterized in that: The device comprises: Acquisition module, used to obtain historical standard contract data; A model training module, used for fine-tuning the pre-trained large language model according to the historical standard contract data; The keyword extraction module uses a large language model that has been fine-tuned to extract keywords from the electronic contract text; A prompt generation module is used to obtain electronic contract review prompts; The review module is used to analyze the keywords using the electronic contract review prompt to obtain the contract risk status of the electronic contract text.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.
10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.
Citation Information
Cited By
Intelligent contract auditing method and device through semantic analysis
CN119477227A
Intelligent file review and result pre-judgment method based on machine learning and application
CN119990104A
Contract auditing method and device based on large model, equipment and medium
CN120317837A
Contract generation method and system based on multi-round dialogue interaction and intelligent element extraction
CN120611727A
Contract generation method and system based on multi-round dialogue interaction and intelligent element extraction
CN120611727B
