Contract evaluation method, system and equipment based on large model and medium

Abstract

The invention discloses a contract evaluation method, system and device based on a large model and a medium, and the method comprises the steps: obtaining a target contract, carrying out the data preprocessing, obtaining the preprocessed contract data, and inputting the preprocessed contract data into a preset contract large language model; the preset contract large language model stores a vector knowledge base including coastal bulk cargo transportation knowledge, legal clauses and industry regulations; in a preset contract big language model, performing contract classification and contract summary extraction on the current preprocessed contract data by using a first model agent to obtain contract summary information; performing contract risk assessment on the current preprocessed contract data to obtain risk assessment information and / or risk avoidance suggestion information; performing benefit measurement and calculation on the current preprocessed contract by using a third model agent to obtain benefit measurement and calculation information; according to the method, the examination and evaluation of the large model in the coastal bulk cargo transportation contract are realized, and the purposes of saving the labor cost and improving the business turnover rate are achieved.

Description

Technical Field

[0001] The present application relates to the technical field of contract evaluation, and in particular to a large model-based contract evaluation method, system, device and medium. Background Art

[0002] At present, the interpretation of coastal bulk shipping contracts mainly relies on manual work, which takes a lot of time and energy, is inefficient and prone to misunderstandings. In addition, for complex contract terms, it may be limited by personal experience and professional knowledge. Some existing contract management software has relatively simple functions and can only perform simple storage and retrieval of contracts, and cannot deeply analyze the meaning and potential risks of contract terms. Based on this, how to provide a contract evaluation method that can save labor costs and improve business turnover has practical application value and significance. Summary of the invention

[0003] In order to realize the review and evaluation of large models in coastal bulk shipping contracts, save labor costs and improve business turnover, the present application provides a contract evaluation method, system, equipment and medium based on large models.

[0004] In the first aspect, the invention objective of the present application is achieved by adopting the following technical solutions: A contract evaluation method based on a large model, including: Acquire the target contract and perform data preprocessing to obtain preprocessed contract data, and input the preprocessed contract data into a preset contract large language model; the preset contract large language model stores a vector knowledge base including coastal bulk cargo transportation knowledge, legal terms and industry regulations; In the preset large contract language model, the first model agent is used to perform contract classification and contract summary extraction on the current pre-processed contract data to obtain contract summary information; the second model agent and the contract summary information are used to perform contract risk assessment on the current pre-processed contract data to obtain risk assessment information and / or risk avoidance suggestion information; the third model agent is used to perform benefit calculation on the current pre-processed contract to obtain benefit calculation information.

[0005] By adopting the above technical scheme, the target contract is the contract data of the coastal bulk cargo transportation contract, which includes the coastal bulk cargo transportation contract whose original contract is paper material and electronic material; the preset contract big language model is a pre-trained big language model; wherein the first model agent is a contract classification model agent, the second model agent is a risk assessment model agent, and the third model agent is a benefit measurement model agent; the contract big language model provided by the present application realizes the review and evaluation of the big model in the coastal bulk cargo transportation contract, so as to save manpower costs and improve business turnover rate; compared with the traditional manual interpretation method, the automated data processing and contract data interpretation process of the present invention is conducive to saving a lot of manpower costs and time costs, greatly improving the efficiency and accuracy of contract interpretation, and reducing misreading and omissions caused by human factors; through the pre-trained contract big language model, not only the data preprocessing operation of the contract data can be realized, but also the contract summary information extraction, contract risk assessment and contract benefit measurement can be realized, which is conducive to improving the operation efficiency of the enterprise.

[0006] In a preferred example of the present application: the target contract includes a paper contract and an electronic contract, including: When the target contract is a paper contract, the paper contract is identified as a plurality of electronic text data using optical character recognition technology, the plurality of electronic text data are preprocessed in sequence, and the plurality of preprocessed electronic text data identified as the same contract are connected into a coherent target contract text using a semantic linking agent; The obtaining of the target contract and performing data preprocessing to obtain preprocessed contract data specifically includes: When the target contract is an electronic contract or a target contract text, a number of corresponding preset prompt words are obtained based on the target contract, and key digital information of the target contract is extracted by using the semantic linking agent in combination with the number of preset prompt words to obtain key contract data; Based on the plurality of preset prompt words and key contract data, data preprocessing is performed on the target contract text and the electronic contract to obtain preprocessed contract data.

[0007] By adopting the above technical solution, the semantic linking agent is an agent based on a large model. This application uses optical character recognition technology (OCR) to identify paper contracts to convert the text in the image into editable electronic text data to form a coherent target contract text, and performs data preprocessing on the target text data and the electronic contract. The data preprocessing includes but is not limited to word segmentation, part-of-speech tagging, named entity recognition, etc. to obtain structured preprocessed contract data. This application uses intelligent contract recognition and processing procedures to shorten the contract evaluation cycle and improve business turnover.

[0008] In a preferred example of the present application, the first model agent is a contract classification agent; in the preset large contract language model, the first model agent is used to classify the current pre-processed contract data and extract the contract summary to obtain the contract summary information, which specifically includes: In the preset contract big language model, the current pre-processed contract data is classified by using the contract classification agent and text classification prompt words to obtain the current contract type; based on the current contract type, the preset contract big language model is optimized, and the optimized contract big language model is used to extract the key elements of the contract from the current pre-processed contract data to obtain the contract summary information.

[0009] By adopting the above technical solutions and introducing the contract classification agent, the type of contract can be identified more accurately. Based on the dynamic optimization of the contract type, the large contract language model can process different types of contracts more accurately, improve the adaptability and generalization ability of the model, and help make more scientific and reasonable risk avoidance recommendations.

[0010] In a preferred example of the present application, the second model agent includes a contract risk assessment agent; the second model agent and the contract summary information are used to perform a contract risk assessment on the current pre-processed contract data to obtain risk assessment information and / or risk avoidance suggestion information, specifically including: In the preset contract language model, based on the current contract type, relevant legal and regulatory information and corresponding requirements and standards are retrieved in the vector knowledge base to obtain risk assessment avoidance information and corresponding assessment standards; The contract risk assessment agent performs a contract risk assessment on the current pre-processed contract data based on the contract summary information, risk assessment avoidance information and corresponding assessment criteria to obtain risk assessment information; and / or, The contract risk assessment agent performs risk point assessment on the current pre-processed contract data based on the contract summary information, risk assessment avoidance information and corresponding assessment standards, and associates the legal provisions and industry regulations based on the several risk points assessed to determine the corresponding risk avoidance recommendation information.

[0011] By adopting the above technical solutions, risk assessment information and risk avoidance advice information can more comprehensively consider the requirements of laws, regulations and industry regulations based on the contract type and contract content of the current pre-processed contract, thereby improving the accuracy and reliability of contract risk assessment. The automated and intelligent contract risk assessment process greatly reduces manual intervention and repetitive work, improves assessment efficiency, and detailed risk assessment information and risk avoidance advice provide strong support for decision makers.

[0012] In a preferred example of the present application, the third model agent includes a benefit calculation agent; the third model agent is used to calculate the benefit of the current pre-processing contract to obtain benefit calculation information, specifically including: In the preset contract large language model, the benefit calculation intelligent agent is used to extract the contract content corresponding to the benefits in the current pre-processed contract data, and the preset benefit calculation calculator is used to perform benefit calculation and evaluation according to the benefit calculation rules corresponding to the current contract type to obtain benefit calculation information.

[0013] By adopting the above technical solutions, the benefit calculation information obtained by the benefit calculation agent through preset benefit calculation rules can more accurately evaluate the expected benefits of the contract and reduce the impact of human errors and subjective judgments; the automated and intelligent benefit calculation process greatly shortens the decision-making cycle and improves decision-making efficiency.

[0014] In a preferred example of the present application, the vector knowledge base also includes industry practices and historical cases of different contract types, and the target contract is obtained and data is preprocessed to obtain preprocessed contract data. Before the preprocessed contract data is input into the preset contract language model, the following is also included: Collect contract texts and related background information of different contract types, and use the initial contract large language model to perform in-depth analysis on the contract text. The in-depth analysis includes performing entity naming recognition on the contract text, extracting key elements of the contract and performing semantic relationship extraction; constructing a logical association diagram between contract elements to form basic contract measurement data; In the initial contract large language model, compliance checking and compliance risk level classification are performed on the contract content of the target contract according to the vector knowledge base to obtain compliance assessment information; Taking the contract basic measurement data, compliance assessment report and credit rating of the contract signatory as inputs of the initial contract large language model, and combining the industry characteristics and regional differences corresponding to the target contract, constructing a feature vector set for contract risk assessment through feature engineering; Based on the feature vector set, the initial contract large language model is trained using a deep learning algorithm and adjusted and optimized to obtain a final contract large language model.

[0015] By adopting the above technical solution, this application can more accurately evaluate the compliance and risk level of the contract through the implementation steps of contract data collection and preprocessing, construction of basic contract measurement data, compliance assessment, feature vector set construction and model training optimization, and perform feature engineering in combination with the industry characteristics and regional differences corresponding to the target contract, so that the contract large language model can adapt to the contract evaluation needs in different scenarios, improve the generalization ability of the model, and through compliance checks and risk level classification, timely discover and correct potential compliance risk points in the contract, thereby reducing the compliance risk of the enterprise.

[0016] In a preferred example of the present application: the in-depth analysis of the contract text further includes: Perform word segmentation on the contract text, remove stop words, and perform word vector representation; Presetting specific prompts in the initial contract large language model to guide the initial contract large language model to understand the contract context; when performing the named entity recognition task, the initial contract large language model identifies key entities and their attributes in the contract; When extracting semantic relationships, the context understanding capability of the initial contract language model is utilized to construct the logical relationships between contract elements and form a structured representation of contract information.

[0017] By adopting the above technical solutions, deep parsing technologies such as word segmentation, removal of stop words and word vector representation can more accurately understand and represent the key information in the contract text; by presetting specific prompts and fine-tuning the initial contract large language model, the contract large language model is more adapted to the specific needs of contract evaluation, improving the adaptability and accuracy of the model, and the automated contract parsing and structured representation process greatly shortens the evaluation cycle and improves the evaluation efficiency. At the same time, the structured contract information representation also facilitates subsequent automated processing and analysis, and is conducive to reducing errors and deviations caused by human factors.

[0018] In the second aspect, the invention objective of the present application is achieved by adopting the following technical solutions: A contract evaluation system based on a large model, the system comprising: A data preprocessing module is used to obtain a target contract and perform data preprocessing to obtain preprocessed contract data, and input the preprocessed contract data into a preset contract language model; the preset contract language model stores a vector knowledge base including coastal bulk cargo transportation knowledge, legal terms and industry regulations; A contract classification module, used to classify the current pre-processed contract data and extract the contract summary using the first model agent in the preset contract language model to obtain contract summary information; The risk assessment and benefit calculation module is used to use the second model intelligent agent and the contract summary information to perform a contract risk assessment on the current pre-processed contract data to obtain risk assessment information and / or risk avoidance suggestion information; and use the third model intelligent agent to perform a benefit calculation on the current pre-processed contract to obtain benefit calculation information.

[0019] By adopting the above technical scheme, the target contract is the contract data of the coastal bulk cargo transportation contract, which includes the coastal bulk cargo transportation contract whose original contract is paper material and electronic material; the preset contract big language model is a pre-trained big language model; wherein the first model agent is a contract classification model agent, the second model agent is a risk assessment model agent, and the third model agent is a benefit measurement model agent; the contract big language model provided by the present application realizes the review and evaluation of the big model in the coastal bulk cargo transportation contract, so as to save manpower costs and improve business turnover rate; compared with the traditional manual interpretation method, the automated data processing and contract data interpretation process of the present invention is conducive to saving a lot of manpower costs and time costs, greatly improving the efficiency and accuracy of contract interpretation, and reducing misreading and omissions caused by human factors; through the pre-trained contract big language model, not only the data preprocessing operation of the contract data can be realized, but also the contract summary information extraction, contract risk assessment and contract benefit measurement can be realized, which is conducive to improving the operation efficiency of the enterprise.

[0020] In the third aspect, the invention objective of the present application is achieved by adopting the following technical solutions: A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned large model-based contract evaluation method when executing the computer program.

[0021] In a fourth aspect, the invention objective of the present application is achieved by adopting the following technical solutions: A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned large-model-based contract evaluation method.

[0022] In summary, the present application includes at least one of the following beneficial technical effects: 1. The target contract is the contract data of the coastal bulk cargo transportation contract, which includes the coastal bulk cargo transportation contract whose original contract is paper material and electronic material; the preset contract big language model is a pre-trained big language model; wherein the first model agent is a contract classification model agent, the second model agent is a risk assessment model agent, and the third model agent is a benefit measurement model agent; the contract big language model provided by the present application realizes the review and evaluation of the big model in the coastal bulk cargo transportation contract, so as to save manpower cost and improve the business turnover rate; compared with the traditional manual interpretation method, the automated data processing and contract data interpretation process of the present invention is conducive to saving a lot of manpower cost and time cost, greatly improving the efficiency and accuracy of contract interpretation, and reducing misreading and omission caused by human factors; through the pre-trained contract big language model, not only the data preprocessing operation of the contract data can be realized, but also the contract summary information extraction, contract risk assessment and contract benefit measurement can be realized, which is conducive to improving the operation efficiency of the enterprise; 2. Through the implementation steps of contract data collection and preprocessing, construction of contract basic measurement data, compliance assessment, feature vector set construction and model training optimization, the compliance and risk level of the contract can be assessed more accurately. Feature engineering is carried out in combination with the industry characteristics and regional differences corresponding to the target contract, so that the contract language model can adapt to the contract assessment needs in different scenarios, improve the generalization ability of the model, and through compliance inspection and risk level classification, timely discover and correct potential compliance risk points in the contract, thereby reducing the compliance risk of the enterprise; 3. By adopting the above technical solutions, deep parsing technologies such as word segmentation, stop word removal and word vector representation can more accurately understand and represent the key information in the contract text; by presetting specific prompts and fine-tuning the initial contract large language model, the contract large language model is more adapted to the specific needs of contract evaluation, improving the adaptability and accuracy of the model, and the automated contract parsing and structured representation process greatly shortens the evaluation cycle and improves the evaluation efficiency. At the same time, the structured contract information representation is also convenient for subsequent automated processing and analysis, and is conducive to reducing errors and deviations caused by human factors. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a flow chart of a contract evaluation method based on a large model in an embodiment of the present application; Figure 2 It is a flowchart before step S1 in a contract evaluation method based on a large model in an embodiment of the present application; Figure 3 It is a schematic diagram of a device in an embodiment of the present application. DETAILED DESCRIPTION

[0024] The present application is further described in detail below in conjunction with the accompanying drawings.

[0025] In one embodiment, if Figure 1 As shown, the present application discloses a contract evaluation method based on a large model, which specifically includes the following steps: S1: Obtain the target contract and perform data preprocessing to obtain preprocessed contract data, and input the preprocessed contract data into a preset contract large language model; the preset contract large language model stores a vector knowledge base including coastal bulk shipping knowledge, legal terms and industry regulations.

[0026] In this embodiment, data preprocessing includes but is not limited to document format conversion, text cleaning, word segmentation and vectorization, part-of-speech tagging, named entity recognition, etc.; wherein the document format conversion converts the contract document into a text format, such as a paper contract, which requires the use of practical optical character recognition technology for conversion, and text cleaning includes removing irrelevant information in the text, such as headers and footers, page numbers, comments, etc., while correcting typos and format errors in the text; word segmentation and vectorization includes word segmentation of the cleaned text, and using a pre-trained word embedding model (such as BERT, GPT, etc.) to convert the word segmentation results into vector representations for subsequent model processing.

[0027] S2: In the preset large contract language model, the first model agent is used to perform contract classification and contract summary extraction on the current pre-processed contract data to obtain contract summary information; the second model agent and the contract summary information are used to perform contract risk assessment on the current pre-processed contract data to obtain risk assessment information and / or risk avoidance suggestion information; the third model agent is used to perform benefit measurement on the current pre-processed contract to obtain benefit measurement information.

[0028] In this embodiment, the classification algorithm is built into the contract language model. The contracts in this embodiment include many different types. For example, the coastal bulk cargo transportation contract includes the long-term cargo consignor contract, the long-term cargo carrier contract, the single-ticket cargo consignor contract and the single-ticket cargo carrier contract. The contract language model of this application is the llama3 model, and it has been trained with the lora of the self-made data set. When defining the Lora module, it must be closely combined with the text characteristics of the coastal bulk cargo transportation contract and the specific requirements of the summary task. Carefully consider at which layers of the model to add trainable parameters in order to most effectively improve the model's ability to capture key information and the quality of summary generation for such contract texts; for example, we can focus on introducing the Lora module in the layers that are closely related to semantic understanding and information extraction to accurately grasp the core content of the contract; model training is to try a relatively small value in the initial stage, such as 1e-4, and then flexibly adjust it according to the performance and convergence of the model during the training process. The optimizer uses the Adam optimizer. When performing model training optimization, by calculating the loss value, the model can clearly understand the gap between itself and the expected output, and use the optimizer to gradually adjust the parameters of the Lora module, thereby continuously improving the ability and quality of summary generation.

[0029] In one embodiment, a contract evaluation method based on a large model includes: S10: When the target contract is a paper contract, the paper contract is identified as multiple electronic text data using optical character recognition technology, the multiple electronic text data are preprocessed in turn, and the multiple preprocessed electronic text data identified as the same contract are connected into a coherent target contract text using a semantic linking agent.

[0030] In this embodiment, a paper contract is scanned using high-precision optical character recognition (OCR) technology to convert the text in the image into editable electronic text data; then the multiple electronic text data obtained by conversion are cleaned, including removing noise, correcting OCR recognition errors, unifying the format, etc., to improve the accuracy of subsequent processing.

[0031] In step S1, the target contract is obtained and data preprocessing is performed to obtain preprocessed contract data, which specifically includes: S11: When the target contract is an electronic contract or a target contract text, obtain a number of corresponding preset prompt words based on the target contract, use the semantic linking intelligent agent in combination with the preset prompt words to extract key digital information of the target contract, and obtain key contract data.

[0032] In this embodiment, the semantic associations among multiple electronic text data are analyzed through a semantic linking agent (agent based on a large model), multiple text fragments belonging to the same contract are identified and merged to form a coherent target contract text.

[0033] S12: Preprocess the target contract text and the electronic contract based on a number of preset prompt words and key contract data to obtain preprocessed contract data.

[0034] In this embodiment, the preset prompt words include prompt words covering the key elements of the contract, including the contract type, signing date, amount and information of both parties, etc.; the semantic linking intelligent agent is combined with the preset prompt words, and the key digital information in the contract (such as amount, date, etc.) is extracted through natural language processing (NLP) technology to form the key data of the contract.

[0035] Specifically, data preprocessing includes but is not limited to word segmentation, part-of-speech tagging, named entity recognition, etc., to obtain structured preprocessed contract data.

[0036] In one embodiment, in step S2, the first model agent is a contract classification agent; in the preset contract large language model, the first model agent is used to perform contract classification and contract summary extraction on the current pre-processed contract data to obtain contract summary information, which specifically includes: S201: In the preset contract big language model, the contract classification agent and text classification prompt words are used to classify the current pre-processed contract data to obtain the current contract type; based on the current contract type, the preset contract big language model is optimized, and the optimized contract big language model is used to extract the key elements of the contract from the current pre-processed contract data to obtain the contract summary information.

[0037] In this embodiment, the contract classification agent is responsible for the contract classification task by preparing or training a set of text classification prompt words for the contract classification agent. These text classification prompt words are used to guide the model to understand and identify different contract types. Then, according to the current contract type obtained by classification, the parameters or weights of the preset contract large language model are dynamically adjusted to make it more focused on processing this type of contract text.

[0038] Specifically, the contract language model automatically extracts key elements in the contract, such as information about the two parties to the contract, description of the subject matter, amount, term, liability for breach of contract, etc., by understanding the semantics and structure of the contract text, forms contract summary information, and outputs the contract summary information in a structured form.

[0039] Furthermore, the optimization process of the contract classification agent and the large language model can be dynamic, that is, the model can continuously learn and improve as new contracts are added and evaluated; a domain knowledge base of the model can also be established, that is, a specialized knowledge base in the contract field can be established to assist the model in understanding and processing professional terms and complex clauses in the contract text.

[0040] In one embodiment, in step S2, the second model agent includes a contract risk assessment agent; the second model agent and the contract summary information are used to perform a contract risk assessment on the current pre-processed contract data to obtain risk assessment information and / or risk avoidance suggestion information, specifically including: S211: In the preset contract language model, relevant legal and regulatory information and corresponding requirements and standards are retrieved in the vector knowledge base based on the current contract type to obtain risk assessment avoidance information and corresponding assessment standards.

[0041] In this embodiment, the vector knowledge base should include rich information such as legal and regulatory texts, judicial interpretations, industry standards, and historical cases, and this information has been processed through vectorization technology to facilitate subsequent rapid matching and query.

[0042] Specifically, legal and regulatory information directly related to the current contract type and corresponding regulations and requirements are screened out from the search results of the vector knowledge base to form a collection of risk assessment avoidance information and assessment standards.

[0043] S212: The contract risk assessment agent performs a contract risk assessment on the current pre-processed contract data based on the contract summary information, risk assessment avoidance information and corresponding assessment criteria to obtain risk assessment information.

[0044] In this embodiment, contract risk assessment includes: inputting contract summary information (including key elements and terms of the contract) into the contract risk assessment intelligent body, the contract risk assessment intelligent body combines risk assessment avoidance information and corresponding assessment criteria to conduct an in-depth analysis of the contract summary information, and then the contract risk assessment intelligent body generates risk assessment information based on the results of the in-depth analysis, including but not limited to risk type, risk level, potential impact, etc.

[0045] and / or, S213: The contract risk assessment agent conducts risk point assessment on the current pre-processed contract data based on the contract summary information, risk assessment avoidance information and corresponding assessment standards, and associates the legal provisions and industry regulations based on the several risk points assessed to determine the corresponding risk avoidance recommendation information.

[0046] In this embodiment, on the basis of risk assessment, the specific risk points in the contract are further evaluated, that is, a detailed interpretation of the contract terms and a comparison with laws, regulations and industry regulations. For several identified risk points, the contract risk assessment agent searches the vector knowledge base for relevant legal provisions and industry regulations, conducts association analysis, and then based on the results of the association analysis, the agent generates risk avoidance suggestion information, including suggestions for modifying contract terms, compliance suggestions, risk response strategies, etc.

[0047] In one embodiment, the third model agent includes a benefit calculation agent; the third model agent is used to calculate the benefit of the current pre-processing contract to obtain benefit calculation information, which specifically includes: S221: In the preset contract language model, the benefit calculation agent is used to extract the contract content corresponding to the benefits in the current pre-processed contract data, and the preset benefit calculation calculator is used to perform benefit calculation and evaluation according to the benefit calculation rules corresponding to the current contract type to obtain benefit calculation information.

[0048] In this embodiment, the benefit calculation agent is trained to recognize and understand key benefit-related content in the contract text, such as the contract amount, payment method, service period, and expected benefits.

[0049] Specifically, the benefit measurement agent conducts an in-depth analysis of the current pre-processed contract data, and uses natural language processing (NLP) technology to extract the contract content that directly corresponds to the benefits; then, based on the current contract type, the corresponding benefit calculation rules are searched in the preset benefit measurement calculator, where the benefit calculation rules may vary due to factors such as contract type, industry characteristics, and regional differences, and it should be ensured that the benefit calculation rules fully match the current contract type in order to perform accurate benefit measurement; then the extracted contract content corresponding to the benefits is input into the preset benefit measurement calculator, and the calculator performs numerical calculations, financial analysis, and other processing on the contract content according to the matching benefit calculation rules to evaluate the expected benefits of the contract and generate benefit measurement information, including but not limited to key indicators such as expected revenue, cost recovery period, and return on investment.

[0050] Furthermore, the benefit measurement agent needs to be trained with a large amount of contract text data to improve its ability to identify benefit-related contract content; the preset benefit calculation rule base needs to be updated and maintained regularly to reflect the impact of market changes, policy adjustments and other factors on benefit measurement, and a flexible rule configuration mechanism should be established to facilitate customized adjustments based on different contract types and customer needs.

[0051] In one embodiment, if Figure 2 As shown, the vector knowledge base also includes industry practices and historical cases of different contract types. Before step S1, a large model-based contract evaluation method also includes: S11: Collect contract texts and related background information of different contract types, and use the initial contract large language model to perform deep analysis on the contract text. The deep analysis includes performing entity naming recognition on the contract text, extracting key elements of the contract and performing semantic relationship extraction; constructing a logical association diagram between contract elements to form basic contract measurement data.

[0052] In this embodiment, the contract texts of different contract types include but are not limited to sales contracts, purchase contracts, lease contracts, service contracts, etc.

[0053] Specifically, the initial contract language model is used to conduct in-depth analysis of the cleaned contract text, including entity naming recognition (identifying the parties in the contract), extracting key elements of the contract (such as contract amount, performance period, breach of contract liability, etc.) and performing semantic relationship extraction (understanding the logical relationship between the elements). The logical association diagram shows the dependency and constraint relationship between the elements, and the logical association diagram is converted into quantifiable basic contract measurement data, such as the frequency of occurrence of key elements, the complexity of semantic relationships, etc., to provide a basis for subsequent evaluation.

[0054] S12: In the initial contract large language model, compliance checks and compliance risk level classification are performed on the contract content of the target contract based on the vector knowledge base to obtain compliance assessment information.

[0055] Specifically, in the initial contract language model, industry practices and historical case information in the vector knowledge base are integrated, and then the contract content of the target contract is checked for compliance, compared with the compliance standards in the vector knowledge base, and potential compliance risk points are identified; then, based on the severity and probability of occurrence of the compliance risk points, the target contract is divided into compliance risk levels and a compliance assessment report is generated.

[0056] S13: The basic contract measurement data, compliance assessment report and credit rating of the contract signatory are used as the input of the initial contract language model. Combined with the industry characteristics and regional differences corresponding to the target contract, a feature vector set for contract risk assessment is constructed through feature engineering.

[0057] Specifically, feature engineering includes steps such as data normalization, feature selection, and feature dimensionality reduction. It uses contract basic measurement data, compliance assessment reports, and credit ratings of contract signatories as input data, and combines the industry characteristics and regional differences corresponding to the target contract to construct a feature vector set for contract risk assessment through feature engineering.

[0058] S14: Based on the feature vector set, the initial contract language model is trained using a deep learning algorithm and adjusted and optimized to obtain the final contract language model.

[0059] Specifically, deep learning algorithms include LSTM and BERT. During the training process, model parameters are continuously adjusted and optimized through cross-validation, hyperparameter tuning and other technical means to improve the accuracy and generalization ability of the model. At the same time, the trained model is evaluated using the test set to verify its performance in practical applications. Based on the evaluation results, the model is further adjusted and optimized until satisfactory performance is achieved.

[0060] In one embodiment, in step S11, the in-depth parsing of the contract text further includes: S111: Perform word segmentation on the contract text, remove stop words, and perform word vector representation.

[0061] In this embodiment, word segmentation processing obtains contract vocabulary units, and stop words existing in the contract vocabulary units after word segmentation processing, such as "de", "le", "zai", etc., are removed to reduce data redundancy and improve processing efficiency.

[0062] Specifically, the vocabulary after word segmentation and removal of stop words is converted into a vector representation in a high-dimensional space, which can be achieved through a pre-trained word embedding model (such as the embedding layer of Word2Vec, GloVe, or BERT) to capture the semantic relationships between words.

[0063] S112: Preset specific prompts in the initial contract large language model to guide the initial contract large language model to understand the contract context; when performing the named entity recognition task, the initial contract large language model recognizes the key entities and their attributes in the contract.

[0064] In this embodiment, the configuration of the initial contract large language model includes presetting specific prompts and model fine-tuning. Among them, presetting specific prompts means presetting a series of specific prompts or guiding words related to contract evaluation in the initial contract large language model to help the model better understand the contract context, such as "Please analyze the key terms in the following contract text", "Please pay attention to identifying the parties and their relationships in the contract", etc.; model fine-tuning includes fine-tuning the initial contract large language model according to the preset prompts and the specific requirements of contract evaluation for different contract types. Model fine-tuning includes re-training some parameters based on a large dataset of contract texts to make the accuracy of the model in the contract evaluation task meet the requirements.

[0065] S113: When performing semantic relationship extraction, utilize the context understanding ability of the initial contract large language model to construct the logical relationships between contract elements and form a structured representation of contract information.

[0066] In this embodiment, the logical relationships include dependency relationships, causal relationships, and parallel relationships between contract elements; relationship extraction includes extracting the semantic relationships between contract elements based on the results of context understanding and representing them in a structured form; the structured representation forms include graph structures or tables of entities, attributes, and relationships, etc., which can intuitively display contract information.

[0067] It should be understood that the sequence numbers of the steps in the above embodiments do not mean the order of execution. The execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0068] In one embodiment, a contract evaluation system based on a big model is provided, and the contract evaluation system based on a big model corresponds to the contract evaluation method based on a big model in the above embodiment.

[0069] A contract evaluation system based on a large model includes a data preprocessing module, a contract classification module, a risk assessment module, and a benefit measurement module. The detailed description of each functional module is as follows: The data preprocessing module is used to obtain the target contract and perform data preprocessing to obtain preprocessed contract data, and input the preprocessed contract data into a preset contract language model; the preset contract language model stores a vector knowledge base including coastal bulk cargo transportation knowledge, legal terms and industry regulations; The contract classification module is used to classify the current pre-processed contract data and extract the contract summary by using the first model agent in the preset contract language model to obtain the contract summary information; The risk assessment and benefit calculation module is used to use the second model intelligent agent and the contract summary information to perform a contract risk assessment on the current pre-processed contract data to obtain risk assessment information and / or risk avoidance suggestion information; and use the third model intelligent agent to perform a benefit calculation on the current pre-processed contract to obtain benefit calculation information.

[0070] For the specific limitations of a contract evaluation system based on a big model, please refer to the limitations of a contract evaluation method based on a big model mentioned above, which will not be repeated here; each module in the above-mentioned contract evaluation system based on a big model can be implemented in whole or in part through software, hardware and a combination thereof; each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the corresponding operations of each of the above modules.

[0071] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 3As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, 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 a large language model of the contract, contract summary information, risk assessment information and benefit measurement information, etc. The network 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, a contract evaluation method based on a large model is implemented.

[0072] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of a large model-based contract evaluation method when executing the computer program.

[0073] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of a contract evaluation method based on a large model are implemented.

[0074] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented 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 memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0075] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0076] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, it should be understood by those skilled in the art that the technical solutions described in the aforementioned embodiments may still be modified, or some of the features thereof may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A contract evaluation method based on a large model, characterized in that: include: Acquire the target contract and perform data preprocessing to obtain preprocessed contract data, and input the preprocessed contract data into a preset contract large language model; the preset contract large language model stores a vector knowledge base including coastal bulk cargo transportation knowledge, legal terms and industry regulations; In the preset large contract language model, the first model agent is used to perform contract classification and contract summary extraction on the current pre-processed contract data to obtain contract summary information; Using the second model agent and the contract summary information, perform contract risk assessment on the current pre-processed contract data to obtain risk assessment information and / or risk avoidance suggestion information; The third model agent is used to calculate the benefit of the current pre-processing contract to obtain benefit calculation information.

2. A contract evaluation method based on a large model according to claim 1, characterized in that: The target contracts include paper contracts and electronic contracts, including: When the target contract is a paper contract, the paper contract is identified as a plurality of electronic text data using optical character recognition technology, the plurality of electronic text data are preprocessed in sequence, and the plurality of preprocessed electronic text data identified as the same contract are connected into a coherent target contract text using a semantic linking agent; The obtaining of the target contract and performing data preprocessing to obtain preprocessed contract data specifically includes: When the target contract is an electronic contract or a target contract text, a number of corresponding preset prompt words are obtained based on the target contract, and key digital information of the target contract is extracted by using the semantic linking agent in combination with the number of preset prompt words to obtain key contract data; Based on the plurality of preset prompt words and key contract data, data preprocessing is performed on the target contract text and the electronic contract to obtain preprocessed contract data.

3. The contract evaluation method based on a large model according to claim 1 is characterized in that: The first model agent is a contract classification agent; in the preset contract language model, the first model agent is used to classify the current pre-processed contract data and extract the contract summary to obtain the contract summary information, which specifically includes: In the preset contract big language model, the current pre-processed contract data is classified by using the contract classification agent and text classification prompt words to obtain the current contract type; based on the current contract type, the preset contract big language model is optimized, and the optimized contract big language model is used to extract the key elements of the contract from the current pre-processed contract data to obtain the contract summary information.

4. A contract evaluation method based on a large model according to claim 3, characterized in that: The second model agent includes a contract risk assessment agent; the method of using the second model agent and the contract summary information to perform a contract risk assessment on the current pre-processed contract data to obtain risk assessment information and / or risk avoidance suggestion information specifically includes: In the preset contract language model, based on the current contract type, relevant legal and regulatory information and corresponding requirements and standards are retrieved in the vector knowledge base to obtain risk assessment avoidance information and corresponding assessment standards; The contract risk assessment agent performs a contract risk assessment on the current pre-processed contract data based on the contract summary information, risk assessment avoidance information and corresponding assessment criteria to obtain risk assessment information; and / or, The contract risk assessment agent performs risk point assessment on the current pre-processed contract data based on the contract summary information, risk assessment avoidance information and corresponding assessment standards, and associates the legal provisions and industry regulations based on the several risk points assessed to determine the corresponding risk avoidance recommendation information.

5. A contract evaluation method based on a large model according to claim 3, characterized in that: The third model agent includes a benefit measurement agent; The third model agent is used to calculate the benefit of the current pre-processing contract to obtain benefit calculation information, which specifically includes: In the preset contract large language model, the benefit calculation intelligent agent is used to extract the contract content corresponding to the benefits in the current pre-processed contract data, and the preset benefit calculation calculator is used to perform benefit calculation and evaluation according to the benefit calculation rules corresponding to the current contract type to obtain benefit calculation information.

6. A contract evaluation method based on a large model according to claim 1, characterized in that: The vector knowledge base also includes industry practices and historical cases of different contract types. The acquisition of the target contract and data preprocessing to obtain preprocessed contract data, before inputting the preprocessed contract data into the preset contract language model, also includes: Collect contract texts and related background information of different contract types, and use the initial contract large language model to perform in-depth analysis on the contract text. The in-depth analysis includes performing entity naming recognition on the contract text, extracting key elements of the contract and performing semantic relationship extraction; constructing a logical association diagram between contract elements to form basic contract measurement data; In the initial contract large language model, compliance checking and compliance risk level classification are performed on the contract content of the target contract according to the vector knowledge base to obtain compliance assessment information; Taking the contract basic measurement data, compliance assessment report and credit rating of the contract signatory as inputs of the initial contract large language model, and combining the industry characteristics and regional differences corresponding to the target contract, constructing a feature vector set for contract risk assessment through feature engineering; Based on the feature vector set, the initial contract large language model is trained using a deep learning algorithm and adjusted and optimized to obtain a final contract large language model.

7. A contract evaluation method based on a large model according to claim 6, characterized in that: The in-depth analysis of the contract text also includes: Perform word segmentation on the contract text, remove stop words, and perform word vector representation; Presetting specific prompts in the initial contract large language model to guide the initial contract large language model to understand the contract context; when performing the named entity recognition task, the initial contract large language model identifies key entities and their attributes in the contract; When extracting semantic relationships, the context understanding capability of the initial contract language model is utilized to construct the logical relationships between contract elements and form a structured representation of contract information.

8. A contract evaluation system based on a large model, characterized in that: The system comprises: A data preprocessing module is used to obtain a target contract and perform data preprocessing to obtain preprocessed contract data, and input the preprocessed contract data into a preset contract language model; the preset contract language model stores a vector knowledge base including coastal bulk cargo transportation knowledge, legal terms and industry regulations; A contract classification module, used to classify the current pre-processed contract data and extract the contract summary using the first model agent in the preset contract language model to obtain contract summary information; The risk assessment and benefit calculation module is used to use the second model intelligent agent and the contract summary information to perform a contract risk assessment on the current pre-processed contract data to obtain risk assessment information and / or risk avoidance suggestion information; and use the third model intelligent agent to perform a benefit calculation on the current pre-processed contract to obtain benefit calculation information.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the large model-based contract evaluation method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of a large model-based contract evaluation method as described in any one of claims 1 to 7 are implemented.

Images