Manufacturing industry contract review method, device and equipment based on large model and medium

Abstract

The invention discloses a manufacturing industry contract review method, device and equipment based on a large model, and a medium, and relates to the technical field of artificial intelligence, and the method comprises the steps: converting a target paper contract image into a to-be-processed contract text through employing an optical character recognition technology, executing a preset data cleaning operation and a preset structured processing operation on the to-be-processed contract text based on a natural language processing technology to obtain a target contract text; dividing the target contract text into model training data, model verification data and model test data to train a preset contract evaluation large model, and optimizing the target parameters to obtain a target preset contract evaluation large model; and obtaining a target entity and a target entity relationship by using the target preset contract evaluation large model to determine a target contract element, and executing a preset compliance evaluation operation and a preset risk evaluation operation on the target contract element to obtain a target evaluation result so as to complete contract evaluation. In this way, automatic and intelligent contract review at a higher level is achieved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and particularly to a method, device, equipment and medium for manufacturing contract review based on large models. Background Art

[0002] In the traditional manufacturing industry, contract review is a crucial and complex task, which is not only related to the legality and compliance of contracts, but also involves the assessment and control of contract risks. Since contract texts usually contain a large amount of professional content such as technical parameters, delivery conditions and quality standards, there are extremely high requirements for the accuracy and efficiency of review. However, traditional manufacturing contract review methods based on large models rely on manual reading and understanding, which is not only time-consuming but also error-prone, making it difficult to ensure the consistency and accuracy of review results. With the development of the manufacturing industry, the increasing number and complexity of contracts make traditional review methods increasingly difficult to meet the industry's needs.

[0003] In recent years, large model technologies have made remarkable progress in the field of natural language processing, especially in text understanding and information extraction. These technologies can understand and process complex language structures by deep learning a large amount of text data, showing great potential in scenarios such as contract review. The introduction of large model technologies is expected to improve the quality and efficiency of manufacturing contract review. These technologies can understand and analyze contract texts through deep learning and natural language processing, identify key terms and potential risks, thereby improving review efficiency and accuracy. Nevertheless, existing large model technologies still have limitations in specific application scenarios in the manufacturing industry, such as insufficient understanding of industry-specific terms and processes, and limited ability to identify industry-specific risk points.

[0004] In summary, how to achieve a higher level of automated and intelligent contract review is an urgent problem to be solved at present. Summary of the Invention

[0005] In view of this, the purpose of the present invention is to provide a method, device, equipment and medium for manufacturing contract review based on large models, which can achieve a higher level of automated and intelligent contract review. The specific solutions are as follows:

[0006] In the first aspect, the present application provides a method for manufacturing contract review based on large models, including:

[0007] Using optical character recognition technology to convert the target paper contract image into a contract text to be processed in a target text format, and performing a preset data cleaning operation and a preset structured processing operation on the contract text to be processed based on natural language processing technology to obtain a target contract text; the target paper contract image is an image obtained by imaging a target manufacturing contract;

[0008] Divide the obtained target contract text into model training data, model verification data, and model test data. Train a preset contract evaluation large model based on the model training data, optimize the target parameters in the preset contract evaluation large model using the model verification data, and then evaluate the optimized preset contract evaluation large model according to the model test data to obtain a target preset contract evaluation large model;

[0009] Use the target preset contract evaluation large model to obtain each target entity in the contract to be reviewed and the target entity relationships between the target entities, and determine the target contract elements in the contract to be reviewed according to the target entities and the target entity relationships. Perform a preset compliance evaluation operation and a preset risk evaluation operation on the target contract elements to obtain a target review result, so as to complete the contract review.

[0010] Optionally, the use of optical character recognition technology to convert the target paper contract image into a contract text to be processed in a target text format includes:

[0011] Perform a preset image denoising operation on the target paper contract image to obtain a denoised contract image;

[0012] Use an adaptive contrast enhancement algorithm to perform a preset image contrast enhancement operation on the denoised contract image to obtain a contract image with enhanced contrast;

[0013] Identify the text area in the contract image with enhanced contrast through a preset text area detection operation;

[0014] Segment the identified text area based on a projection analysis algorithm or a connected component analysis method to obtain corresponding target characters;

[0015] Use an OCR recognition engine based on deep learning to extract features and classify the target characters, so as to convert the target characters into structured text data;

[0016] Perform a preset text recombination operation on the structured text data according to the layout logic of the text area to obtain a contract text to be processed in a target text format that conforms to the preset coding specification.

[0017] Optionally, the preset data cleaning operation and preset structured processing operation performed on the contract text to be processed based on natural language processing technology include:

[0018] Perform a preset data cleaning operation on the contract text to be processed based on natural language processing technology, and then convert the contract text to be processed that meets the preset unstructured format conditions into a target contract text that meets the preset structured format conditions according to the natural language processing technology;

[0019] Among them, the preset data cleaning operation includes any one or more of a preset format error correction operation, a preset duplicate data deletion operation, and a preset text error correction operation.

[0020] Optionally, the obtaining of each target entity in the contract to be reviewed and the target entity relationship between the target entities by using the target preset contract evaluation large model includes:

[0021] Performing a preset multimodal parsing operation on the contract to be reviewed to generate a parsed contract to be reviewed that meets the preset specification conditions;

[0022] Obtaining each target entity in the parsed contract to be reviewed by using the target preset contract evaluation large model; among them, the target entity includes any one or more of a subject entity, a clause entity, and a numerical entity;

[0023] Establishing a cross-segment entity association rule for the target entities in the parsed contract to be reviewed, so as to determine the target entity relationship between the target entities based on the cross-segment entity association rule.

[0024] Optionally, the performing of a preset compliance evaluation operation and a preset risk evaluation operation on the target contract elements to obtain a target review result includes:

[0025] Checking the key clauses in the contract to be reviewed based on the preset review rules and the target contract elements;

[0026] Generating a compliance report corresponding to the contract to be reviewed according to the obtained inspection result;

[0027] Comparing the key clauses with preset standard clauses to determine the target risk points corresponding to the key clauses;

[0028] Performing a preset classification operation and a preset scoring operation on the target risk points based on a preset risk scoring range to determine the target risk level of the target risk points;

[0029] Generating a corresponding target review result according to the target risk level of the target risk points and the compliance report.

[0030] Optionally, after performing a preset compliance evaluation operation and a preset risk evaluation operation on the target contract elements to obtain a target review result, it further includes:

[0031] Comparing the contract to be reviewed with a preset standard contract template and / or the historical version of the contract to be reviewed to determine the contract category and content change of the contract to be reviewed;

[0032] Display the content changes of the contract to be reviewed based on the preset visual prompt operation;

[0033] Store the contract to be reviewed in a preset contract database according to the contract category of the contract to be reviewed.

[0034] Optionally, after performing the preset compliance assessment operation and the preset risk assessment operation on the target contract elements to obtain the target review result, it further includes:

[0035] Dynamically map the target review result to a preset visualization template to determine the target visualization template corresponding to the target review result based on the preset visualization template;

[0036] Based on a preset coding conversion algorithm, convert the numerical indicators, risk level classification data, and clause association relationships in the target review result into visualization elements in the target visualization template;

[0037] Generate a corresponding visualization result using a dynamic rendering algorithm according to the target visualization template and the visualization elements;

[0038] Wherein, the target visualization template includes any one or more of a preset chart template, a preset dashboard template, and a preset heat map template.

[0039] In a second aspect, the present application provides a manufacturing contract review device based on a large model, including:

[0040] A text acquisition module, configured to use optical character recognition technology to convert a target paper contract image into a contract text to be processed in a target text format, and perform a preset data cleaning operation and a preset structured processing operation on the contract text to be processed based on natural language processing technology to obtain a target contract text; the target paper contract image is an image obtained by imaging a target manufacturing contract;

[0041] A model training module, configured to divide the obtained target contract text into model training data, model verification data, and model test data, train a preset contract evaluation large model based on the model training data, optimize the target parameters in the preset contract evaluation large model using the model verification data, and then evaluate the optimized preset contract evaluation large model according to the model test data to obtain a target preset contract evaluation large model;

[0042] A review result acquisition module, configured to use the target preset contract evaluation large model to obtain each target entity in the contract to be reviewed and the target entity relationships between the target entities, determine the target contract elements in the contract to be reviewed according to the target entities and the target entity relationships, and perform a preset compliance evaluation operation and a preset risk evaluation operation on the target contract elements to obtain a target review result, so as to complete contract review.

[0043] Thirdly, the present application provides an electronic device, including:

[0044] A memory, configured to store a computer program;

[0045] A processor, configured to execute the computer program to implement the manufacturing contract review method based on a large model as described above.

[0046] Fourthly, the present application provides a computer-readable storage medium, configured to store a computer program; wherein, when the computer program is executed by a processor, the manufacturing contract review method based on a large model as described above is implemented.

[0047] In summary, in this application, the target paper contract image is first converted into a contract text to be processed in a target text format by using optical character recognition technology, and a preset data cleaning operation and a preset structuring operation are performed on the contract text to be processed based on natural language processing technology to obtain a target contract text; the target paper contract image is an image obtained by imaging a target manufacturing contract; the obtained target contract text is divided into model training data, model verification data, and model test data, a preset contract evaluation large model is trained based on the model training data, the target parameters in the preset contract evaluation large model are optimized by using the model verification data, and then the optimized preset contract evaluation large model is evaluated according to the model test data to obtain a target preset contract evaluation large model; the target entities in the contract to be reviewed and the target entity relationships between the target entities are obtained by using the target preset contract evaluation large model, and the target contract elements in the contract to be reviewed are determined according to the target entities and the target entity relationships, and a preset compliance evaluation operation and a preset risk evaluation operation are performed on the target contract elements to obtain a target review result, so as to complete contract review. As can be seen from the above, in this application, the target paper contract image obtained by imaging the target manufacturing contract is first converted into a contract text to be processed in a target text format by using optical character recognition technology, and then a preset data cleaning and structuring operation is performed on it based on natural language processing technology to obtain a target contract text; then the target contract text is divided into model training data, model verification data, and model test data, a preset contract evaluation large model is trained based on the model training data, the target parameters therein are optimized by using the model verification data, and the target preset contract evaluation large model is obtained by evaluating according to the model test data; finally, the target entities and entity relationships in the contract to be reviewed are obtained by using this model, the target contract elements are determined accordingly, and a preset compliance evaluation and risk evaluation operation are performed on the elements to obtain a target review result, completing contract review. In this way, this application realizes the automatic extraction, review, comparison, and archiving management of contract content by introducing advanced large model artificial intelligence technology. In addition, through an intelligent early warning mechanism, risk points in the contract are identified in advance to help enterprises take effective risk control measures, thereby reducing contract risks. Description of the Drawings

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only embodiments of the present invention, and for those of ordinary skill in the art, other drawings can be obtained based on the provided drawings without creative efforts.

[0049] Figure 1 It is a flowchart of a method for reviewing manufacturing contracts based on a large model disclosed in this application;

[0050] Figure 2 Structural schematic diagram of a manufacturing contract review device based on a large model disclosed in this application;

[0051] Figure 3 Structural diagram of an electronic device disclosed in this application. Detailed implementation manners

[0052] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0053] Currently, in recent years, large model technologies have made remarkable progress in the field of natural language processing, especially in text understanding and information extraction. These technologies can understand and process complex language structures by deep learning a large amount of text data, showing great potential in scenarios such as contract review. The introduction of large model technologies is expected to improve the quality and efficiency of contract review in the manufacturing industry. These technologies can understand and analyze contract texts through deep learning and natural language processing, identify key terms and potential risks, thereby improving review efficiency and accuracy. Nevertheless, existing large model technologies still have limitations in specific application scenarios in the manufacturing industry, such as insufficient understanding of industry-specific terms and processes, and limited ability to identify industry-specific risk points. To solve the above technical problems, this application discloses a manufacturing contract review method, device, equipment and medium based on a large model, which can achieve a higher level of automated and intelligent contract review.

[0054] See Figure 1 As shown, an embodiment of the present invention discloses a manufacturing contract review method based on a large model, including:

[0055] Step S11: Use optical character recognition technology to convert the target paper contract image into a contract text to be processed in a target text format, and perform a preset data cleaning operation and a preset structured processing operation on the contract text to be processed based on natural language processing technology to obtain a target contract text; the target paper contract image is an image obtained by imaging a target manufacturing contract.

[0056] In this embodiment, first, a preset image denoising operation needs to be performed on the target paper contract image to obtain the denoised contract image; the adaptive contrast enhancement algorithm is used to perform a preset image contrast enhancement operation on the denoised contract image to obtain the contract image with enhanced contrast; the text area in the contract image with enhanced contrast is recognized through a preset text area detection operation; the recognized text area is segmented based on the projection analysis algorithm or the connected region analysis method to obtain the corresponding target characters; the OCR (Optical Character Recognition) recognition engine based on deep learning is used to extract and classify the features of the target characters, so as to convert the target characters into structured text data; the structured text data is subjected to a preset text recombination operation according to the layout logic of the text area to obtain the contract text to be processed in a target text format that conforms to the preset coding specification. Specifically, the optical character recognition technology is used to perform a preset image denoising operation on the target paper contract image in the form of PDF (Portable Document Format), scanned pictures, etc. to obtain the denoised contract image. Then, the contrast of the denoised contract image is enhanced to obtain the contract image with enhanced contrast. At the same time, the text area in the contract image with enhanced contrast is recognized, and non-contract main body contents such as headers, footers, watermarks, and annotations are removed. Then, the target characters in the text area are obtained, and the target characters are converted into structured text data. Finally, the structured text data is subjected to a preset text recombination operation according to the original layout logic of the target paper contract image to obtain the contract text to be processed.

[0057] In this embodiment, a preset data cleaning operation is performed on the contract text to be processed based on natural language processing technology, and then the contract text to be processed that meets the preset unstructured format conditions is converted into a target contract text that meets the preset structured format conditions according to the natural language processing technology; wherein, the preset data cleaning operation includes any one or several of a preset format error correction operation, a preset duplicate data deletion operation, and a preset text error correction operation. Specifically, the NLP (Natural Language Processing) technology and the rule engine are combined to automatically identify and process various data problems. For example, format errors are corrected, duplicate data is deleted, and obvious errors in the text are corrected. Next, entities in the contract can be recognized, such as personal or organizational names, locations, dates, etc. The relationship extraction technology can be used to recognize the relationships between entities, such as the contract relationship between "Party A" and "Party B". The text classification technology is used to recognize the type or theme of the contract, and the clause classification technology is used to recognize and mark specific clauses in the contract, such as payment clauses, liability for breach of contract, confidentiality agreements, etc. These entities are converted into a unified preset structured format to obtain the target contract text.

[0058] Step S12: Divide the obtained target contract text into model training data, model validation data, and model test data. Train a preset contract evaluation large model based on the model training data, optimize the target parameters in the preset contract evaluation large model using the model validation data, and then evaluate the optimized preset contract evaluation large model according to the model test data to obtain a target preset contract evaluation large model.

[0059] In this embodiment, target contract texts are obtained from different sources, and these target contract texts include, but are not limited to, various types of commercial contracts, legal documents, etc. The obtained target contract text is used as training data, and the training data is divided into a model training set, a model validation set, and a model test set. The preset contract evaluation large model first learns on the model training set and then tunes on the model validation set to prevent overfitting. Finally, the performance of the preset contract evaluation large model on the model test set will be used to evaluate its generalization ability. If the preset contract evaluation large model meets the preset requirements after training, it is determined as the target preset contract evaluation large model. Additionally, during the training process, the selection and adjustment of hyperparameters are involved, such as the learning rate, batch size, number of network layers, etc. The selection of these hyperparameters has a decisive impact on the performance of the model.

[0060] It should be understood that the ultimate goal of training the preset contract evaluation large model is to accurately understand and analyze the contract content. The preset contract evaluation large model should not only be able to identify key information in the contract, such as the contract parties, dates, amounts, etc., but also be able to understand the meaning and legal consequences of the contract terms. Therefore, various technologies and methods, such as transfer learning and reinforcement learning, will be adopted during the model training process to improve the model's understanding and reasoning abilities. Through training based on a large amount of contract text data, the large model can learn the complex features and patterns of the contract text, thereby realizing the automated analysis and processing of the contract content. With the continuous development and optimization of deep learning technologies, the training of the large model will continuously improve the model's performance and accuracy, providing a more intelligent and efficient solution for contract management.

[0061] Step S13: Use the target preset contract evaluation large model to obtain each target entity in the contract to be reviewed and the target entity relationships between the target entities, determine the target contract elements in the contract to be reviewed according to the target entities and the target entity relationships, and perform a preset compliance evaluation operation and a preset risk evaluation operation on the target contract elements to obtain a target review result, so as to complete the contract review.

[0062] In this embodiment, it is necessary to perform a preset multimodal parsing operation on the contract to be reviewed to generate a parsed contract to be reviewed that meets the preset specification conditions; use the target preset contract evaluation large model to obtain each target entity in the parsed contract to be reviewed; where the target entity includes any one or several of the subject entity, clause entity, and numerical entity; establish a cross-segment implementation entity association rule for the target entity in the parsed contract to be reviewed, so as to determine the target entity relationship between the target entities based on the cross-segment implementation entity association rule. Specifically, the contract to be reviewed may contain complex legal terms, industry-specific expressions, and diverse format layouts. Therefore, it is necessary to perform a preset multimodal parsing operation on the contract to be reviewed according to different contract types and requirements of the contract to be reviewed, and obtain the parsed contract to be reviewed. Then, identify the entities, concepts, and relationships in the parsed contract to be reviewed through the target preset contract evaluation large model, so as to accurately extract the target key elements in the contract to be reviewed. For example, important information such as the first party and the second party in the contract, the contract signing date, payment terms, liability for breach of contract, etc. can be identified.

[0063] Furthermore, based on the preset review rules and the target contract elements, check the key clauses in the contract to be reviewed; generate a compliance report corresponding to the contract to be reviewed according to the obtained inspection results; compare the key clauses with the preset standard clauses to determine the target risk points corresponding to the key clauses; perform a preset classification operation and a preset scoring operation on the target risk points based on the preset risk scoring range to determine the target risk level of the target risk points; generate a corresponding target review result according to the target risk level of the target risk points and the compliance report. Specifically, based on a deep understanding of legal clauses and compliance requirements, convert the provisions of laws and regulations into a series of executable preset review rules, and check the key clauses in the contract to be reviewed based on the preset review rules and the target contract elements. For example, the price clause in the contract needs to comply with relevant price laws and regulations, the delivery clause needs to meet industry standards, and the confidentiality clause needs to follow data protection regulations. Furthermore, obtain the corresponding inspection results and generate a corresponding compliance report. As laws and regulations are revised and industry standards are adjusted, the preset review rules will also change accordingly. Therefore, it has a certain degree of flexibility and scalability to be able to update and adjust the preset rules in a timely manner.

[0064] In addition, data analysis techniques and machine learning algorithms can be used to perform pre-set risk assessment operations on various terms and conditions in the contract to be reviewed, so as to identify risk points that may have an adverse impact on the enterprise. Through comprehensive analysis of the contract text, potential legal, financial, and operational risks are revealed, thereby helping the enterprise take preventive measures. For example, check whether there is a clear performance guarantee in the contract; whether there are terms that may result in one party assuming unreasonable responsibilities; whether there are payment terms that may lead to unexpected financial losses. In addition, the vague or unclear expressions in the contract can be evaluated. Through comprehensive analysis of the target contract elements, a comprehensive risk profile can be provided for the enterprise, pointing out areas that require special attention.

[0065] In this embodiment, the contract to be reviewed is compared with a pre-set standard contract template and / or the historical version of the contract to be reviewed to determine the contract category and content changes of the contract to be reviewed; the content changes of the contract to be reviewed are displayed based on pre-set visual prompt operations; and the contract to be reviewed is stored in a pre-set contract database according to the contract category of the contract to be reviewed. Specifically, load the contract draft and the referenced standard contract template or historical contract, and then use a text comparison algorithm to identify the differences between the two. This includes not only the addition and deletion of words, but also changes in the structure, order, and logical relationship of the terms, to ensure that the comparison result is not only at the literal level, but can deeply understand the semantics and legal meaning of the contract terms. For example, identify terms with the same meaning even though the wording is different, and expressions that are seemingly the same but actually have differences. Secondly, while identifying the differences, the possible impacts of these differences will also be evaluated. At the same time, it can also be combined with the enterprise's compliance database and risk management strategy to automatically detect terms in the draft that may violate laws, regulations, or company policies. This proactive risk identification mechanism helps the enterprise adjust and optimize the contract content in a timely manner before contract signing, avoiding potential disputes and losses in the future. Finally, the content changes of the contract to be reviewed are displayed to the user based on pre-set visual prompt operations, and the user can quickly browse the differences between the contract draft and the standard template through this interface. In addition, use a pre-set database or content management to store contract documents and provide search and retrieval functions so that users can quickly find the required contracts.

[0066] In this embodiment, the target review result is dynamically mapped to a preset visualization template to determine the target visualization template corresponding to the target review result based on the preset visualization template; the numerical indicators, risk level classification data, and clause association relationships in the target review result are converted into visualization elements in the target visualization template based on a preset coding conversion algorithm; a corresponding visualization result is generated using a dynamic rendering algorithm according to the target visualization template and the visualization elements; wherein, the target visualization template includes any one or several of a preset chart template, a preset dashboard template, and a preset heat map template. Specifically, a preset visualization template is determined through a customized visualization tool, such as a preset chart template, a preset dashboard template, and a preset heat map template, etc. These templates are dynamically mapped to the target review result, and the data information in the target review result is imported into the selected target visualization template. In addition, personalized settings can be provided, allowing the display content and format to be adjusted according to requirements, such as the display of data points, chart types, and color schemes. This integration ability improves the efficiency and effectiveness of contract management. In this way, through an intuitive and easy-to-understand display, the user experience is improved, and it is continuously optimized with the progress of technology to meet the user's needs for efficient and intelligent contract review.

[0067] As can be seen from the above, in the embodiment of the present application, first, the target paper contract image obtained by imaging the target manufacturing contract is converted into a contract text to be processed in a target text format through optical character recognition technology, and then a preset data cleaning and structuring operation is performed on it based on natural language processing technology to obtain a target contract text; then the target contract text is divided into model training data, model verification data, and model test data, a preset contract evaluation large model is trained based on the model training data, the target parameters in it are optimized using the model verification data, and the target preset contract evaluation large model is obtained according to the evaluation of the model test data; finally, the target entities and entity relationships in the contract to be reviewed are obtained using this model, the target contract elements are determined accordingly, and a preset compliance evaluation and risk assessment operation is performed on the elements to obtain a target review result, completing the contract review. In this way, in the embodiment of the present application, by introducing advanced large model artificial intelligence technology, the automated extraction, review, comparison, and archiving management of contract content are realized. In addition, through an intelligent early warning mechanism, risk points in the contract are identified in advance to help enterprises take effective risk control measures, thereby reducing contract risks.

[0068] Based on the previous embodiment, it can be known that the present application discloses a method for manufacturing contract review based on a large model, which can achieve a higher level of automated and intelligent contract review. Next, a method for manufacturing contract review based on a large model will be described in detail.

[0069] This application first uses optical character recognition technology to perform a preset image denoising operation on the target paper contract image in the form of PDF, scanned pictures, etc., to obtain the denoised contract image. Then, the contrast of the denoised contract image is enhanced to obtain the contract image with enhanced contrast. At the same time, the text area in the contract image with enhanced contrast is recognized, and then the target characters in the text area are obtained. The target characters are converted into structured text data, and the structured text data is subjected to a preset text recombination operation according to the original layout logic of the target paper contract image, and finally the contract text to be processed is obtained. Combining NLP technology and rule engines to automatically identify and process various data problems, for example, correcting format errors, deleting duplicate data, and correcting obvious errors in the text. Next, the entities in the contract can be identified and these entities can be converted into a unified preset structured format to obtain the target contract text.

[0070] Secondly, obtain the target contract texts from different sources and use them as training data, and divide the training data into a model training set, a model validation set, and a model test set. The preset contract evaluation large model first learns on the model training set and then tunes on the model validation set to prevent overfitting. Finally, the performance of the preset contract evaluation large model on the model test set will be used to evaluate its generalization ability. If the preset contract evaluation large model meets the preset requirements after training, it is determined as the target preset contract evaluation large model.

[0071] Finally, through the target preset contract evaluation large model, identify and analyze the target entities, concepts, and target entity relationships in the contract to be reviewed, so as to accurately extract the target key elements in the contract to be reviewed. Subsequently, based on the preset review rules and the target contract elements, check the key terms in the contract to be reviewed to obtain a compliance report. At the same time, perform a preset risk assessment operation on various terms and conditions in the contract to be reviewed according to the obtained target key elements to identify the risk points that may have an adverse impact on the enterprise. Obtain the target review result according to the compliance report and the risk points.

[0072] In this way, this application can use the manufacturing contract review large model to analyze the content of manufacturing contracts and identify potential legal risks and non-compliant terms.

[0073] See Figure 2 As shown, an embodiment of the present invention discloses a manufacturing contract review device based on a large model, which may include:

[0074] A text acquisition module 11, configured to convert a target paper contract image into a to-be-processed contract text in a target text format by using optical character recognition technology, and perform a preset data cleaning operation and a preset structuring operation on the to-be-processed contract text based on natural language processing technology to obtain a target contract text; the target paper contract image is an image obtained by imaging a target manufacturing contract.

[0075] A model training module 12, configured to divide the obtained target contract text into model training data, model verification data, and model test data, train a preset contract evaluation large model based on the model training data, optimize target parameters in the preset contract evaluation large model by using the model verification data, and then evaluate the optimized preset contract evaluation large model according to the model test data to obtain a target preset contract evaluation large model.

[0076] A review result acquisition module 13, configured to use the target preset contract evaluation large model to obtain each target entity in a to-be-reviewed contract and the target entity relationship between the target entities, determine target contract elements in the to-be-reviewed contract according to the target entities and the target entity relationship, and perform a preset compliance evaluation operation and a preset risk evaluation operation on the target contract elements to obtain a target review result, so as to complete contract review.

[0077] As can be seen from the above, in this application, first, the target paper contract image obtained by imaging the target manufacturing contract is converted into a to-be-processed contract text in a target text format by using optical character recognition technology, and then a preset data cleaning and structuring operation is performed on it based on natural language processing technology to obtain a target contract text; then the target contract text is divided into model training data, model verification data, and model test data, a preset contract evaluation large model is trained based on the model training data, the target parameters in it are optimized by using the model verification data, and a target preset contract evaluation large model is obtained by evaluating according to the model test data; finally, the target entities and entity relationships in the to-be-reviewed contract are obtained by using this model, the target contract elements are determined accordingly, and a preset compliance evaluation and risk evaluation operation are performed on the elements to obtain a target review result, completing contract review. In this way, this application realizes the automatic extraction, review, comparison, and archiving management of contract content by introducing advanced large model artificial intelligence technology. In addition, through an intelligent early warning mechanism, risk points in the contract are identified in advance to help enterprises take effective risk control measures, thereby reducing contract risks.

[0078] In some specific implementation manners, the text acquisition module 11 may specifically include:

[0079] A denoised contract image acquisition unit, configured to perform a preset image denoising operation on the target paper contract image to obtain a denoised contract image.

[0080] A post - contrast - enhancement contract image acquisition unit, which is used to perform a preset image contrast enhancement operation on the denoised contract image by using an adaptive contrast enhancement algorithm to obtain a post - contrast - enhancement contract image;

[0081] A text area recognition unit, which is used to recognize the text area in the post - contrast - enhancement contract image through a preset text area detection operation;

[0082] A target character acquisition unit, which is used to segment the recognized text area based on a projection analysis algorithm or a connected - region analysis method to obtain corresponding target characters;

[0083] A structured text data acquisition unit, which is used to extract features and classify the target characters by using an OCR recognition engine based on deep learning, so as to convert the target characters into structured text data;

[0084] A contract text to be processed acquisition unit, which is used to perform a preset text recombination operation on the structured text data according to the layout logic of the text area to obtain a contract text to be processed in a target text format that conforms to a preset coding specification.

[0085] In some specific embodiments, the text acquisition module 11 may specifically include:

[0086] A target contract text acquisition unit, which is used to perform a preset data cleaning operation on the contract text to be processed based on natural language processing technology, and then convert the contract text to be processed that meets the preset unstructured format conditions into a target contract text that meets the preset structured format conditions according to the natural language processing technology; wherein, the preset data cleaning operation includes any one or several of a preset operation for correcting format errors, a preset operation for deleting duplicate data, and a preset operation for correcting text errors.

[0087] In some specific embodiments, the review result acquisition module 13 may specifically include:

[0088] A parsed contract to be reviewed generation unit, which is used to perform a preset multi - modal parsing operation on the contract to be reviewed to generate a parsed contract to be reviewed that meets the preset specification conditions;

[0089] A target entity acquisition unit, which is used to obtain each target entity in the parsed contract to be reviewed by using the target preset contract evaluation large model; wherein, the target entity includes any one or several of a subject entity, a clause entity, and a numerical entity;

[0090] A target entity relationship determination unit, configured to establish a cross-segment implementation entity association rule for the target entities in the contract to be reviewed after parsing, so as to determine the target entity relationship between the target entities based on the cross-segment implementation entity association rule.

[0091] In some specific embodiments, the review result acquisition module 13 may specifically include:

[0092] A key clause detection unit, configured to check the key clauses in the contract to be reviewed based on a preset review rule and the target contract elements;

[0093] A compliance report generation unit, configured to generate a compliance report corresponding to the contract to be reviewed according to the obtained inspection result;

[0094] A target risk point determination unit, configured to compare the key clauses with preset standard clauses to determine the target risk points corresponding to the key clauses;

[0095] A target risk level determination unit, configured to perform a preset classification operation and a preset scoring operation on the target risk points based on a preset risk scoring range to determine the target risk level of the target risk points;

[0096] A target review result generation unit, configured to generate a corresponding target review result according to the target risk level of the target risk points and the compliance report.

[0097] In some specific embodiments, the manufacturing contract review device based on a large model may further include:

[0098] A contract comparison module, configured to compare the contract to be reviewed with a preset standard contract template and / or a historical version of the contract to be reviewed to determine the contract category and content changes of the contract to be reviewed;

[0099] A content change display module, configured to display the content changes of the contract to be reviewed based on a preset visual prompt operation;

[0100] A contract storage module, configured to store the contract to be reviewed in a preset contract database according to the contract category of the contract to be reviewed.

[0101] In some specific embodiments, the manufacturing contract review device based on a large model may further include:

[0102] A target visualization template determination module, configured to perform dynamic mapping of the target review result with a preset visualization template to determine a target visualization template corresponding to the target review result based on the preset visualization template;

[0103] A visualization element conversion module, configured to convert the numerical metrics, risk level classification data, and clause association relationships in the target review result into visualization elements in the target visualization template based on a preset encoding conversion algorithm;

[0104] A visualization result generation module, configured to generate a corresponding visualization result according to the target visualization template and the visualization elements by using a dynamic rendering algorithm; wherein, the target visualization template includes any one or more of a preset chart template, a preset dashboard template, and a preset heat map template.

[0105] Furthermore, an embodiment of the present application also discloses an electronic device, Figure 3 which is a structural diagram of an electronic device 20 shown according to an exemplary embodiment. The content in the figure should not be regarded as any limitation on the scope of use of the present application.

[0106] Figure 3 This is a schematic structural diagram of an electronic device 20 provided by an embodiment of the present application. The electronic device 20 may specifically include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. Among them, the memory 22 is used to store a computer program, and the computer program is loaded and executed by the processor 21 to implement the relevant steps in the manufacturing contract review method based on a large model disclosed in any of the foregoing embodiments. In addition, the electronic device 20 in this embodiment may specifically be an electronic computer.

[0107] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows is any communication protocol applicable to the technical solution of the present application, and specific limitations are not imposed here; the input / output interface 25 is used to obtain external input data or output data to the outside, and its specific interface type can be selected according to specific application needs, and specific limitations are not imposed here.

[0108] In addition, as a carrier for resource storage, the memory 22 may be a read-only memory, a random access memory, a magnetic disk, or an optical disc, etc. The resources stored thereon may include an operating system 221, a computer program 222, etc., and the storage method may be short-term storage or permanent storage.

[0109] Among them, the operating system 221 is used to manage and control each hardware device and computer program 222 on the electronic device 20, and it can be Windows Server, Netware, Unix, Linux, etc. In addition to the computer program that can be used to complete the large model-based manufacturing contract review method executed by the electronic device 20 disclosed in any of the foregoing embodiments, the computer program 222 may further include computer programs that can be used to complete other specific tasks.

[0110] Furthermore, the present application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the foregoing disclosed large model-based manufacturing contract review method. For the specific steps of this method, reference can be made to the corresponding content disclosed in the foregoing embodiments, and details will not be repeated here.

[0111] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the same or similar parts among the various embodiments, reference can be made to each other. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and reference can be made to the description of the method part for related parts.

[0112] Those skilled in the art can further realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed in this article can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the components and steps of each example have been generally described according to their functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0113] The steps of the method or algorithm described in combination with the embodiments disclosed in this article can be directly implemented by hardware, a software module executed by a processor, or a combination of the two. The software module can be placed in a random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, register, hard disk, removable disk, CD-ROM, or any other form of storage medium well-known in the technical field.

[0114] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0115] The technical solution provided by the present application is introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea. At the same time, for general technicians in this field, according to the idea of ​​the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A method for reviewing manufacturing contracts based on large models, characterized in that, Including: Using optical character recognition technology to convert the target paper contract image into a to-be-processed contract text in a target text format, and performing a preset data cleaning operation and a preset structuring operation on the to-be-processed contract text based on natural language processing technology to obtain a target contract text; the target paper contract image is an image obtained by imaging a target manufacturing contract; Dividing the obtained target contract text into model training data, model verification data, and model test data, training a preset contract evaluation large model based on the model training data, optimizing target parameters in the preset contract evaluation large model using the model verification data, and then evaluating the optimized preset contract evaluation large model according to the model test data to obtain a target preset contract evaluation large model; Using the target preset contract evaluation large model to obtain each target entity in the to-be-reviewed contract and the target entity relationship between the target entities, and determining the target contract elements in the to-be-reviewed contract according to the target entities and the target entity relationship, and performing a preset compliance evaluation operation and a preset risk evaluation operation on the target contract elements to obtain a target review result, so as to complete the contract review.

2. The method for reviewing manufacturing contracts based on large models according to claim 1, wherein The step of using optical character recognition technology to convert the target paper contract image into a to-be-processed contract text in a target text format includes: Performing a preset image denoising operation on the target paper contract image to obtain a denoised contract image; Using an adaptive contrast enhancement algorithm to perform a preset image contrast enhancement operation on the denoised contract image to obtain a contract image with enhanced contrast; Identifying the text area in the contract image with enhanced contrast through a preset text area detection operation; Segmenting the identified text area based on a projection analysis algorithm or a connected component analysis method to obtain corresponding target characters; Using a deep learning-based OCR recognition engine to extract features and classify the target characters, so as to convert the target characters into structured text data; Performing a preset text recombination operation on the structured text data according to the layout logic of the text area to obtain a to-be-processed contract text in a target text format that conforms to the preset coding specification.

3. The method for reviewing manufacturing contracts based on large models according to claim 1, wherein The step of performing a preset data cleaning operation and a preset structuring operation on the to-be-processed contract text based on natural language processing technology includes: Performing a preset data cleaning operation on the to-be-processed contract text based on natural language processing technology, and then converting the to-be-processed contract text that meets the preset unstructured format condition into a target contract text that meets the preset structured format condition according to the natural language processing technology; Among them, the preset data cleaning operation includes any one or several of a preset format error correction operation, a preset duplicate data deletion operation, and a preset text error correction operation.

4. The method for reviewing manufacturing contracts based on large models according to claim 1, wherein The step of using the target preset contract evaluation large model to obtain each target entity in the to-be-reviewed contract and the target entity relationship between the target entities includes: Performing a preset multimodal parsing operation on the to-be-reviewed contract to generate a parsed to-be-reviewed contract that meets the preset specification conditions; Use the target preset contract evaluation large model to obtain each target entity in the parsed contract to be reviewed; wherein, the target entity includes any one or several of a subject entity, a clause entity, and a numerical entity; Establish cross-segment implementation entity association rules for the target entities in the parsed contract to be reviewed, so as to determine the target entity relationships between the target entities based on the cross-segment implementation entity association rules.

5. The method for reviewing manufacturing contracts based on large models according to any one of claims 1 to 4, characterized in that, The performing the preset compliance evaluation operation and the preset risk evaluation operation on the target contract elements to obtain a target review result includes: Check the key clauses in the contract to be reviewed based on the preset review rules and the target contract elements; Generate a compliance report corresponding to the contract to be reviewed according to the obtained inspection result; Compare the key clauses with the preset standard clauses to determine the target risk points corresponding to the key clauses; Perform a preset classification operation and a preset scoring operation on the target risk points based on the preset risk scoring range to determine the target risk level of the target risk points; Generate a corresponding target review result according to the target risk level of the target risk points and the compliance report.

6. The method for reviewing manufacturing contracts based on large models according to claim 5, characterized in that, After performing the preset compliance evaluation operation and the preset risk evaluation operation on the target contract elements to obtain a target review result, it further includes: Compare the contract to be reviewed with the preset standard contract template and / or the historical version of the contract to be reviewed to determine the contract category and content changes of the contract to be reviewed; Display the content changes of the contract to be reviewed based on the preset visual prompt operation; Store the contract to be reviewed in a preset contract database according to the contract category of the contract to be reviewed.

7. The method for reviewing manufacturing contracts based on a large model according to claim 6, wherein After performing the preset compliance evaluation operation and the preset risk evaluation operation on the target contract elements to obtain a target review result, it further includes: Dynamically map the target review result to a preset visualization template to determine a target visualization template corresponding to the target review result based on the preset visualization template; Convert the numerical indicators, risk level classification data, and clause association relationships in the target review result into visualization elements in the target visualization template based on a preset coding conversion algorithm; Generate a corresponding visualization result according to the target visualization template and the visualization elements using a dynamic rendering algorithm; Wherein, the target visualization template includes any one or several of a preset chart template, a preset dashboard template, and a preset heat map template.

8. A manufacturing contract review device based on a large model, characterized in that, It includes: A text acquisition module, configured to use optical character recognition technology to convert a target paper contract image into a contract text to be processed in a target text format, and perform a preset data cleaning operation and a preset structured processing operation on the contract text to be processed based on natural language processing technology to obtain a target contract text; the target paper contract image is an image obtained by imaging a target manufacturing contract; A model training module, configured to divide the obtained target contract text into model training data, model validation data, and model test data, train a preset contract evaluation large model based on the model training data, optimize target parameters in the preset contract evaluation large model by using the model validation data, and then evaluate the optimized preset contract evaluation large model according to the model test data to obtain a target preset contract evaluation large model; A review result acquisition module, configured to use the target preset contract evaluation large model to obtain each target entity in the contract to be reviewed and the target entity relationships between the target entities, determine target contract elements in the contract to be reviewed according to the target entities and the target entity relationships, and perform a preset compliance evaluation operation and a preset risk evaluation operation on the target contract elements to obtain a target review result, so as to complete contract review.

9. An electronic device, characterized in that, It includes: A memory, configured to store a computer program; A processor, configured to execute the computer program to implement the large model-based manufacturing contract review method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, For storing a computer program; wherein, when the computer program is executed by the processor, it implements the large model-based manufacturing contract review method according to any one of claims 1 to 7.

Images