A large model-based contract intelligent review method and system

Abstract

The application discloses a kind of contract intelligent examination method and system based on large model, which comprises the following steps: obtaining the contract text needing examination;Obtain the input examination requirement;According to the examination requirement, the contract text is processed by comprehensively using morphological analysis, syntax analysis, semantic analysis and discourse analysis to extract key information and contract terms;According to the extracted key information and contract terms, the contract text is analyzed based on large model, and favorable factors, unfavorable factors and processing suggestions are identified;The identified favorable factors, unfavorable factors and processing suggestions are summarized to generate an analysis report.The present application significantly improves the comprehensiveness and accuracy of contract examination by using large-scale pre-training models and comprehensive multi-domain data, and enhances the understanding and risk assessment capability of novel contract terms.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence legal technology, and in particular to a method and system for intelligent contract review based on a large model. Background Technology

[0002] As businesses continue to grow, the demand for contract review increases dramatically. One of the biggest challenges facing businesses is how to promptly and effectively identify and mitigate hidden risks within contracts during the review process.

[0003] Compared to traditional methods relying on human experience or AI algorithms for inspection, large-scale models offer a completely new technological solution. AI algorithms require ample training data upfront, and to ensure their continued accuracy and efficiency, they necessitate ongoing maintenance by dedicated algorithm engineers—a time-consuming and labor-intensive process. Large-scale models, on the other hand, learn from massive datasets, gaining the ability to recognize complex patterns and relationships. Furthermore, they self-correct and optimize through data observation and analysis, eliminating the need for manually specified rules or features, thus adapting more easily to diverse and complex review tasks.

[0004] In the future, with the rapid development of large-scale modeling technology, it will be more widely applied and gradually change enterprise operation methods, driving productivity improvement. Existing traditional NLP models analyze contract document text primarily through word segmentation, part-of-speech tagging, and syntactic analysis to extract keywords, entities, and relationships from the contract. Then, statistical analysis is used to identify risks within the contract. While this approach can perform basic text processing and information extraction tasks, existing technologies still have some shortcomings and limitations, such as:

[0005] If the training data is insufficient or fails to cover all possible contract terms and expressions, the model may miss some key terms or misinterpret the contract content, affecting the accuracy of the analysis.

[0006] Because traditional models are often customized for specific tasks or domains, they may lack generalization ability when dealing with unseen contract types or novel clauses, making it difficult to accurately understand and assess the risks involved. Summary of the Invention

[0007] In view of this, embodiments of the present invention provide a method and system for intelligent contract review based on a large model to solve the above-mentioned technical problems.

[0008] To achieve the above objectives, firstly, a contract intelligent review method based on a large model is provided, which includes the following steps:

[0009] Obtain the contract text that needs to be reviewed;

[0010] Obtain the review requirements for the input;

[0011] Based on the aforementioned review requirements, the contract text is processed using a combination of lexical analysis, syntactic analysis, semantic analysis, and discourse analysis to extract key information and contract terms.

[0012] Based on the extracted key information and contract terms, the contract text is analyzed for advantages and disadvantages using a large model to identify favorable factors, unfavorable factors, and handling suggestions.

[0013] The identified favorable and unfavorable factors and proposed solutions are summarized to generate an analysis report.

[0014] In some possible implementations, the process of processing the contract text using a combination of lexical analysis, syntactic analysis, semantic analysis, and discourse analysis, based on the review requirements, to extract key information and contract terms, specifically includes:

[0015] During lexical analysis, each word in the contract text is identified based on a word segmentation algorithm, and proper nouns, legal terms, and specific expressions are marked.

[0016] During the syntactic analysis, the syntactic structure of the contract text is analyzed based on the syntactic parsing algorithm to determine the syntactic relationships between words;

[0017] During semantic analysis, semantic understanding algorithms are used to determine the precise meaning of words and phrases in the contract text within the legal context.

[0018] When performing discourse analysis, the discourse analysis algorithm determines the interaction and constraint relationships between different contract clauses based on the context of the contract text.

[0019] In some possible implementations,

[0020] The word segmentation algorithms include: Jieba word segmentation algorithm, Tsinghua University Chinese word segmentation algorithm, or Harbin Institute of Technology Language Technology Platform word segmentation algorithm;

[0021] The syntactic parsing algorithms mentioned include: the Stanford parser or the Harbin Institute of Technology Language Technology Platform syntactic parser;

[0022] The semantic understanding algorithms include: the BERT model based on bidirectional encoder representation or the GPT model of generative pre-trained transformer;

[0023] The aforementioned text analysis algorithms include: TextRank or Latent Dirichlet Allocation (LDA).

[0024] In some possible implementations, the step of performing a strengths and weaknesses analysis on the contract text based on the extracted key information and contract terms, using a large model to identify favorable factors, unfavorable factors, and processing suggestions, specifically includes:

[0025] By combining extracted key information, laws and regulations, historical cases, and contract terms, and based on natural language processing algorithms and legal knowledge graphs, the favorable and unfavorable factors of the contract text are determined.

[0026] Based on the aforementioned favorable factors, the rights and obligations of the contracting parties are listed using an information extraction algorithm, and the product quality, service standards, and acceptance standards are clearly defined.

[0027] Based on the aforementioned unfavorable factors, and in conjunction with the extracted key information and contract terms, proposed modifications are suggested.

[0028] In some possible implementations, the proposed modifications, based on the extracted key information and contract terms, address unfavorable factors and include:

[0029] The algorithm uses a rule engine to identify liability clauses in contracts and combines them with a legal knowledge graph to provide specific suggestions on liability allocation.

[0030] The fuzzy matching algorithm is used to detect ambiguous terms in contracts, and the semantic analysis algorithm is used to provide clear quantitative standards or specific time limits.

[0031] Secondly, a contract intelligent review system based on a large model is provided, which includes:

[0032] The contract text acquisition module is used to acquire the contract texts that need to be reviewed;

[0033] The review requirement acquisition module is used to acquire the input review requirements;

[0034] The contract text processing module is used to process the contract text by comprehensively applying lexical analysis, syntactic analysis, semantic analysis and discourse analysis according to the review requirements, so as to extract key information and contract terms;

[0035] The contract advantages and disadvantages analysis module is used to perform an advantages and disadvantages analysis on the contract text based on the extracted key information and contract terms, and to identify favorable factors, unfavorable factors and handling suggestions based on a large model.

[0036] The analysis report generation module is used to summarize the identified favorable factors, unfavorable factors, and handling suggestions to generate an analysis report.

[0037] In some possible implementations, the contract text processing module specifically includes:

[0038] The lexical analysis submodule is used to identify each word in the contract text based on the word segmentation algorithm during lexical analysis, and to mark proper nouns, legal terms and specific expressions.

[0039] The syntax analysis submodule is used to parse the syntactic structure of the contract text based on the syntax parsing algorithm and determine the syntactic relationships between words during syntax analysis.

[0040] The semantic analysis submodule is used to determine the precise meaning of words and phrases in the contract text in the legal context based on semantic understanding algorithms when performing semantic analysis.

[0041] The document analysis submodule is used to determine the interaction and constraint relationships between different contract clauses based on the context of the contract text during document analysis, using a document analysis algorithm.

[0042] In some possible implementations, the word segmentation algorithm includes: Jieba word segmentation algorithm, Tsinghua University Chinese word segmentation algorithm, or Harbin Institute of Technology language technology platform word segmentation algorithm;

[0043] The syntactic parsing algorithms mentioned include: the Stanford parser or the Harbin Institute of Technology Language Technology Platform syntactic parser;

[0044] The semantic understanding algorithms include: the BERT model based on bidirectional encoder representation or the GPT model of generative pre-trained transformer;

[0045] The aforementioned text analysis algorithms include: TextRank or Latent Dirichlet Allocation (LDA).

[0046] In some possible implementations, the contract advantages and disadvantages analysis module specifically includes:

[0047] The Contract Advantages and Disadvantages Determination Submodule is used to combine extracted key information, laws and regulations, historical cases, and contract clauses, and based on natural language processing algorithms and legal knowledge graphs, to determine the favorable and unfavorable factors of the contract text.

[0048] The favorable factors suggestion generation submodule is used to list the rights and obligations of the contracting parties based on the information extraction algorithm, and to clarify the product quality, service standards and acceptance standards for the aforementioned favorable factors.

[0049] The "Adverse Factors Modification Suggestion" submodule is used to propose modification suggestions for the aforementioned adverse factors, based on extracted key information and contract terms.

[0050] In some possible implementations, the adverse factor modification suggestion submodule specifically includes:

[0051] The liability allocation suggestion unit is used to identify liability clauses in contract terms through rule engine algorithms and provide specific liability allocation suggestions in conjunction with legal knowledge graphs;

[0052] The fuzzy term processing unit is used to detect fuzzy terms in contracts through fuzzy matching algorithms and to provide clear quantitative standards or specific time limit suggestions using semantic analysis algorithms.

[0053] Thirdly, an electronic device is provided, comprising:

[0054] One or more processors;

[0055] Storage device for storing one or more programs.

[0056] When the one or more programs are executed by the one or more processors, the one or more processors implement a large-model-based intelligent contract review method as described in the first aspect.

[0057] Fourthly, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements a large-model-based intelligent contract review method as described in the first aspect.

[0058] Fifthly, a computer program product is provided, comprising a computer-readable storage medium on which a computer program is stored, wherein when the computer program is run, it executes the intelligent contract review method based on a large model described in the first aspect.

[0059] The above technical solution has the following beneficial technical effects:

[0060] This invention automates the acquisition of contract texts and review requirements, and utilizes natural language processing techniques (including lexical analysis, syntactic analysis, semantic analysis, and discourse analysis) to process the contract texts. This method significantly improves the speed of contract review. This automation reduces the need for manual reading and analysis of large amounts of text, thereby accelerating the review process.

[0061] This invention utilizes advanced large-scale modeling techniques, such as transformer-based models (e.g., BERT or GPT), to more accurately understand the complex semantics and structure of contract text. This not only improves the accuracy of information extraction but also ensures the consistency of review results, reducing the impact of human error and subjective judgment.

[0062] This invention, through comprehensive analysis of extracted key information and contract terms, can identify in detail the favorable and unfavorable factors of a contract. This in-depth analysis helps users fully understand the potential risks and benefits of a contract.

[0063] This invention, based on identified disadvantages, can propose specific modification suggestions. These suggestions, grounded in in-depth legal and semantic analysis, can help users optimize contract terms, reduce risks, and enhance the legal validity of contracts.

[0064] The final analysis report generated by this invention summarizes all favorable and unfavorable factors and proposed solutions, providing users with a clear and easy-to-understand review results document. This report can be directly used for further review, modification, and decision support. Attached Figure Description

[0065] The accompanying drawings are provided to better understand the invention and are not intended to unduly limit the scope of the invention. Wherein:

[0066] Figure 1 This is a flowchart of a contract intelligent review method based on a large model, according to an embodiment of the present invention.

[0067] Figure 2 This is a detailed flowchart of step S130 in an embodiment of the present invention;

[0068] Figure 3 This is a detailed flowchart of step S140 in an embodiment of the present invention;

[0069] Figure 4 This embodiment provides a structural block diagram of a contract intelligent review system based on a large model;

[0070] Figure 5 This is a schematic diagram of the structure of a computer system according to an embodiment of the present invention. Detailed Implementation

[0071] The following description, in conjunction with the accompanying drawings, illustrates exemplary embodiments of the present invention, including various details to aid understanding. These details should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the invention. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0072] This invention employs a large-scale review model to analyze the main text of contracts. This model originates from a large-scale pre-trained model (such as BERT, GPT series, or Legal-BERT, which are specifically designed for training on legal texts). Based on the original model, it integrates a large amount of legal literature, legal clauses, contract texts, and templates to construct the knowledge base data source for the review model. This data is then used to train the model, generating the review model. After extensive training, the model possesses a deep understanding and precise analytical capability of legal corpora, enabling it to learn to identify and understand key elements and clauses in contracts, and to determine their compliance, risk level, etc., based on the contract content.

[0073] This model eliminates the need for manual annotation and allows for analysis of contract content by incorporating contextual semantics. After analysis, it outputs suggested modifications and relevant legal basis, ultimately generating documents and reports. This significantly reduces upfront costs and improves the efficiency of subsequent contract processing.

[0074] Example 1

[0075] like Figure 1 As shown, this embodiment provides a contract intelligent review method based on a large model, which is implemented through the following detailed steps:

[0076] S110: Obtain the contract text that needs to be reviewed.

[0077] This step begins by obtaining the contract text to be reviewed. Users upload the contract file through a secure web interface; this file can be in PDF, Word, or other electronic document formats. A built-in file parsing module (such as using the Apache Tika library) is responsible for parsing the uploaded file and extracting the plain text content for subsequent processing. For example, if a user uploads a PDF file of a commercial lease agreement, it will be automatically recognized and converted into a processable text format.

[0078] S120: Requirements for reviewing input.

[0079] In this step, users specify their review requirements within the same web interface. These requirements can be specified by selecting preset review templates, such as "Risk Assessment," "Legality Check," or "Fairness Analysis of Terms," ​​or by entering specific keywords or phrases, such as "Termination Conditions" or "Liability for Indemnification." These inputs are used as parameters for subsequent text analysis and processing. For example, if a user selects the "Risk Assessment" template, they will focus on potentially hidden risk clauses in the contract.

[0080] S130: Process the contract text by comprehensively applying lexical analysis, syntactic analysis, semantic analysis and discourse analysis according to the review requirements, so as to extract key information and contract clauses.

[0081] In this step, advanced natural language processing technology can be used to conduct a comprehensive analysis of the contract text. First, the lexical analyzer decomposes words and phrases in the text and identifies their parts of speech (nouns, verbs, etc.). Next, the syntactic analyzer constructs a grammatical structure tree of the sentence to help understand the relationship between various components. Semantic analysis further interprets the meaning of the sentence to ensure that the true intention of the clause is understood. Discourse analysis evaluates the logical flow and coherence in the text, and identifies the document structure and key parts. For example, the phrase "have no choice but to" identified through semantic analysis expresses a mandatory obligation.

[0082] As Figure 2 shown, in some embodiments, the step of processing the contract text by comprehensively applying lexical analysis, syntactic analysis, semantic analysis and discourse analysis according to the review requirements as described in step S130 to extract key information and contract clauses specifically includes:

[0083] S131: When performing lexical analysis, identify each word in the contract text based on a word segmentation processing algorithm, and mark proper nouns, legal terms and specific expressions.

[0084] The specific execution steps of the lexical analysis in this step are as follows:

[0085] First, text preprocessing is performed to clean the text to remove non-text elements such as images, tables, headers and footers, and standardize the text format, for example, unify line breaks and spaces. Next, word segmentation processing is performed, and the cleaned text is segmented using the Jieba word segmentation algorithm. By constructing a dictionary based on the Trie tree structure, full-mode matching is performed on the text to identify all possible words. Then, part-of-speech tagging is performed on the word segmentation results to identify nouns, verbs, adjectives, etc., with special attention to legal proper nouns and terms, such as "contract", "breach of contract", etc., which are marked using a dedicated legal lexicon. Finally, specific legal expressions in the contract, such as "force majeure", "disclaimer", etc., are identified and marked, and regular expressions and pattern matching technologies are used to identify specific legal phrases and clauses.

[0086] Jieba word segmentation is an efficient word segmentation algorithm implemented based on a prefix dictionary, which is suitable for Chinese text. For example, for the sentence "Party A shall pay the rent on time" in the contract text, Jieba word segmentation decomposes it into "Party A", "shall", "on time", "pay", "rent". In addition, proper nouns, legal terms such as "Party A", "rent" and specific expressions such as "shall" in the word segmentation results are marked to facilitate subsequent analysis.

[0087] S132: During the syntactic analysis, the syntactic structure of the contract text is analyzed based on the syntactic parsing algorithm to determine the syntactic relationships between words.

[0088] Specifically, during the grammatical analysis phase, the Stanford parser can be used to parse the syntactic structure of the contract text. The Stanford parser is a tool capable of recognizing language structure and outputting dependency trees. For example, for the segmented sentence above, the parser will establish dependency relationships between "payment" as the active word, "Party A" (subject), and "rent" (object). This helps the system understand the grammatical function of each word in the sentence.

[0089] The specific steps of syntax analysis in this step are as follows:

[0090] First, dependency parsing is performed by inputting the lexically analyzed text into the Stanford parser. For example, for the sentence "Party A must pay all outstanding payments before the contract terminates," the parser identifies "pay" as the core verb, marks "Party A" as the subject, "payments" as the direct object, "must" as a modal verb, "outstanding payments" as an attributive of payments, and "all" as a quantity modifier of payments. Based on the identified dependency relations, the parser constructs a dependency tree. This tree structure reveals the deep grammatical structure of the sentence, clarifying the role and relationships of each word in the sentence.

[0091] Next, syntactic structure analysis is performed. Based on the dependency tree, the sentence structure is further analyzed to identify main clauses and subordinate clauses, as well as the connections between them. For example, in the sentence "If Party A fails to perform the contract, Party B shall have the right to terminate the contract," the system identifies the conditional clause introduced by "if" and the result clause introduced by "then." The system also identifies and analyzes complex sentence structures such as conditional sentences, causal sentences, and alternative sentences, which are often crucial to the interpretation of the contract. Finally, logical connectors in the sentence, such as "and," "or," and "unless," are identified. These words are often used in contracts to define the scope of rights and obligations. For example, in the sentence "Party A shall provide the equipment and be responsible for its maintenance," "and" connects two independent but related obligations.

[0092] S133: When performing semantic analysis, the precise meaning of words and phrases in the contract text in the legal context is determined based on a semantic understanding algorithm.

[0093] Specifically, in the semantic analysis step, the BERT model can be used to determine the precise meaning of words and phrases in the contract text within the legal context. Through pre-training and deep semantic understanding of the context, the BERT model can accurately grasp the subtle differences in legal texts. For example, the model can distinguish the specific meaning and consequences of "breach of contract" in different legal clauses.

[0094] In the semantic analysis process, the BERT model is first used for contextual understanding. Through its pre-trained context embedding capabilities, the BERT model can deeply understand the specific meaning of each word in the text, especially in complex legal contexts. For example, in a contract text stating "Party A has the right to terminate the contract due to force majeure," the BERT model can understand from the context that "termination" does not simply mean ending, but rather refers to the termination of the contract due to a specific circumstance (force majeure). The same word in different contexts, such as "Party B has failed to fulfill its obligations, and Party A may terminate the contract," indicates that "termination" is due to the breach of contract. BERT's bidirectional training structure allows it to extract information from both preceding and following text, thus more accurately capturing these subtle semantic differences.

[0095] Next, the semantic disambiguation step is performed. Many terms in legal texts may have different meanings in different clauses or contexts. For example, the term "indemnification" in a contract usually refers to compensation for losses caused by breach of contract, but in an insurance contract it may refer to compensation for losses caused by an accident. Another example is "duration of validity," which refers to the period during which the contract is executed in a general commercial contract, while in a licensing agreement it refers to the time frame for authorized use. Through the BERT model, the system can identify the precise meaning of these terms in a specific contract and ensure that the interpretation of each term matches the specific context of the contract, thereby avoiding misunderstandings and potential legal disputes.

[0096] This in-depth semantic analysis not only improves the accuracy of contract review, but also helps legal professionals better understand and enforce contract terms, ensuring the legal validity and reasonableness of contract enforcement.

[0097] S134: When performing discourse analysis, the interaction and constraint relationships between different contract clauses are determined based on the contextual relationships of the contract text according to the discourse analysis algorithm.

[0098] Specifically, in the text analysis phase, the TextRank algorithm can be used to determine the interactions and constraints between different clauses in a contract text. TextRank is a graph-based ranking algorithm that identifies key parts of the text by constructing a graph of the relationships between keywords and sentences. For example, the TextRank algorithm can identify the logical relationships between liability clauses and compensation clauses in a contract, ensuring consistency and reasonableness among the clauses.

[0099] The following are the specific steps for performing text analysis, explained in detail with concrete examples:

[0100] A relational graph is constructed using the TextRank algorithm. The contract text is processed using TextRank to construct a graph consisting of text units (e.g., keywords or sentences) as nodes and their co-occurrence relationships as edges. For example, consider the contract clauses: "Party A shall pay for the goods within 30 days of signing the contract" and "Party B shall ship the goods within 3 days of receiving full payment." Through TextRank, the system identifies "payment" and "shipment" as key actions and connects these two actions using time nodes ("within 30 days", "within 3 days") to form a logical sequence. The system analyzes the constructed graph, identifying key nodes and edges in the text to reveal the logical relationships between the clauses.

[0101] Examine the logical consistency between contract clauses to ensure there are no contradictory or conflicting terms. For example, if another clause states "Party B shall ship the goods within 5 days of signing the contract," this presents a clear time conflict and logical inconsistency with the aforementioned "Party B shall ship the goods within 3 days of receiving full payment." To further analyze and confirm the relationships and consistency between the subject matter of the clauses, the system will employ the Latent Dirichlet Allocation (LDA) model. LDA is a subject model that can uncover hidden subject structures within textual data. In contract review scenarios, LDA can help identify subjects in different clauses and compare their consistency. For example, through LDA analysis, the system finds that both the "payment clause" and the "shipment clause" fall under the subject of "transaction performance," thus helping to confirm the logical consistency and completeness of these clauses.

[0102] The word segmentation algorithms include: Jieba word segmentation algorithm, Tsinghua University Chinese word segmentation algorithm, or Harbin Institute of Technology language technology platform word segmentation algorithm; the syntactic parsing algorithms include: Stanford parser or Harbin Institute of Technology language technology platform syntactic parser; the semantic understanding algorithms include: BERT (Bidirectional Encoder Representations from Transformers) model based on bidirectional encoder representation or GPT (Generative Pre-trained Transformer) model; the discourse analysis algorithms include: TextRank or Latent Dirichlet Allocation (LDA) algorithm.

[0103] S140: Based on the extracted key information and contract terms, perform a strengths and weaknesses analysis on the contract text using a large model to identify favorable factors, unfavorable factors, and handling suggestions.

[0104] In this step, a well-trained large-scale machine learning model (e.g., a Transformer-based model) is used to synthesize the analysis results from the previous three steps S110 to S130, conducting an in-depth analysis of the advantages and disadvantages of the contract terms. The model assesses the risks and benefits of each clause, identifying potential disadvantages and advantages. For example, it may point out that the restrictions of a certain compensation clause are too harsh and suggest appropriate adjustments to protect the user's interests.

[0105] like Figure 3 As shown, specifically, step S140 includes the following steps:

[0106] S141: Combining the extracted key information, laws and regulations, historical cases, and contract clauses, and based on natural language processing algorithms and legal knowledge graphs, determine the favorable and unfavorable factors of the contract text. The following provides a detailed description of the various sub-steps included in step S141:

[0107] S141-1: Extracting Key Information from Contract Text. First, the system scans the uploaded contract text, using keyword extraction and entity recognition algorithms from natural language processing technology to identify and extract key information. This information includes the names of the contracting parties, the contract amount, the contract term, and payment terms. For example, if the contract text contains the sentence: "Party A (the supplier) and Party B (the customer) hereby enter into this contract for a contract amount of RMB 1 million and a contract term of three years," the system will automatically identify "Party A" and "Party B" as the contracting parties, and extract "RMB 1 million" as the contract amount and "three years" as the contract term.

[0108] S141-2: Combining with relevant laws and regulations. The system compares and analyzes the extracted key information with relevant laws and regulations. This step is accomplished by accessing a built-in legal knowledge base, which contains various legal provisions, regulations, and previous legal interpretations. For example, for the extracted "contract term is three years," the system will check the relevant industry's legal provisions, such as whether there are legal requirements for maximum or minimum contract terms in the construction industry.

[0109] S141-3: Analyzing Historical Cases. Next, the system utilizes the historical case database within the legal knowledge graph to analyze historical cases related to the contract. These cases help the system understand how similar clauses are performed and interpreted in actual legal disputes. For example, the system might find a historical case where a "three-year contract term" clause was ruled invalid because it did not comply with specific industry regulations. This will prompt the system to flag this as a potential disadvantage.

[0110] S141-4: Identifying Favorable and Unfavorable Factors. Finally, the system synthesizes the preceding analysis results and uses more advanced natural language processing algorithms (such as sentiment analysis and text classification) to determine the favorable and unfavorable factors in the text. This includes assessing the clarity, legality, and consistency with the contract's purpose of the clauses. For example, if the contract explicitly states that "Party B has the right to request Party A to provide at least three free maintenance services during the contract period," this would be considered a favorable factor because it provides Party B with a clear right. Conversely, if the contract vaguely mentions that "Party A shall endeavor to provide support," it may be considered an unfavorable factor because "endeavor" is a legally ambiguous expression that could lead to disputes regarding its implementation.

[0111] S142: Based on the aforementioned favorable factors, list the rights and obligations of the contracting parties using an information extraction algorithm, and clarify product quality, service standards, and acceptance criteria. The following provides a detailed description of the various sub-steps included in step S142:

[0112] S142-1: Extracting rights and obligations.

[0113] The system first uses information extraction algorithms to identify and list the specific rights and obligations of the parties from the contract text. This step involves text analysis techniques, such as named entity recognition and relation extraction, to ensure accurate extraction of relevant information from the text. For example, the contract may contain a clause stating, "Party A has the right to require Party B to provide at least three free maintenance services during the contract period." The system will automatically identify and list this clause as Party A's right, while simultaneously identifying "three free maintenance services" as Party B's obligation.

[0114] S142-2: Define Product Quality Standards. Next, the system defines specific quality standards for the products or services mentioned in the contract. This involves semantic analysis of relevant descriptions in the contract text to ensure the specificity and enforceability of the standards. For example, if the contract states, "The products provided by Party B must meet national standard XYZ," the system will parse this standard and confirm what "national standard XYZ" specifically refers to, ensuring that this quality standard is clearly recorded and can be used for subsequent quality control and acceptance.

[0115] S142-3: Setting Service Standards. In this step, the system will define the service standards mentioned in the contract. This includes the specific content of the service, the execution standards of the service, and the time frame for service completion. For example, the contract stipulates: "Party B shall complete the initial installation and commissioning of the equipment within 30 days after the signing of the contract." The system will extract this service standard and clarify the specific requirements and time limits of the service to ensure that these standards are specific and enforceable.

[0116] S142-4: Define Acceptance Criteria. Finally, the system will define the acceptance criteria in the contract, which are used to assess whether the product or service meets the contract requirements. This typically involves a detailed analysis of the contract terms to ensure transparency and impartiality in the acceptance process. For example, the contract terms may state: "Product acceptance will be based on the XYZ acceptance test standard." The system will analyze and confirm the specific content of the XYZ acceptance test standard in detail, ensuring that the acceptance criteria are clear, unambiguous, and applicable to the actual product acceptance process.

[0117] Through the above sub-steps, the system can comprehensively handle the favorable factors in the contract, ensure that the rights and obligations of the contracting parties are clear and unambiguous, and clarify product quality, service and acceptance standards, thereby improving the efficiency and satisfaction of contract implementation.

[0118] S143: In response to the aforementioned adverse factors, and in conjunction with the extracted key information and contract terms, propose modification suggestions.

[0119] Specifically, the proposed modifications, based on extracted key information and contract terms, address unfavorable factors and include:

[0120] S143-1: Identifying liability clauses in contract terms using a rule engine algorithm and providing specific liability allocation suggestions in conjunction with a legal knowledge graph. Specifically, the system first applies a rule engine algorithm to scan the contract text to identify liability clauses. This algorithm uses pre-defined rules, such as keywords (e.g., "liability," "breach of contract," "penalty," etc.) and clause patterns, to label contract content related to liability. For example, if the contract contains the clause: "If Party B fails to complete the equipment installation on time, Party A has the right to claim compensation," the system will identify "failure to complete on time" and "compensation" as liability-related keywords. After identifying the liability clauses, the system further analyzes their specific content to assess their clarity and legality. This step involves in-depth semantic analysis of the clauses to ensure understanding of their specific intent and legal consequences. For example, the system will analyze whether the specific time standard for "completing the equipment installation on time" is clearly stated in the contract, and whether the specific calculation method for "compensation" is reasonable. The system uses a built-in legal knowledge graph, combined with the identified and analyzed liability clauses, to provide specific liability allocation suggestions. The legal knowledge graph includes relevant laws and regulations, historical cases, and interpretations by legal experts, helping the system to make reasonable and legally compliant suggestions. For example, if the "compensation" method in the original clause is not clear enough, the system may suggest modifying it to: "If Party B fails to complete the equipment installation before the date stipulated in the contract, Party B shall pay Party A a penalty equivalent to 5% of the contract amount."

[0121] S143-2: Detect ambiguous terms in contracts using fuzzy matching algorithms and provide clear quantitative standards or specific time limits using semantic analysis algorithms.

[0122] Specifically, when dealing with ambiguous terms in contracts, the system first performs sub-step 1: using a fuzzy matching algorithm to identify ambiguous terms. This algorithm scans the contract text and identifies terms that need further clarification based on a pre-defined list of ambiguous terms (such as "as soon as possible," "appropriately," "when necessary," etc.). For example, if the contract contains the sentence: "Party B shall complete the installation of the equipment as soon as possible," the system will identify "as soon as possible" as an ambiguous term because it does not provide a specific timeframe.

[0123] Next, the system proceeds to sub-step 2: semantic analysis to determine contextual meaning. In this step, the system delves into the text surrounding vague terms to determine their specific meaning within the contract's particular context. This involves analyzing clauses related to the vague terms, such as equipment arrival time and pre-installation preparations, to infer the actual expected time for "as soon as possible."

[0124] Finally, the system executes sub-step 3: providing clear quantitative standards or time limits. Based on the semantic analysis results, the system will propose specific suggestions for replacing vague terms, making the contract terms more explicit and enforceable. For example, regarding "Party B shall complete the installation of the equipment as soon as possible," the system may suggest revising it to: "Party B shall complete the installation of the equipment within 30 days after the signing of the contract." Such a revision not only provides a clear time limit but also helps reduce legal disputes arising from inconsistent interpretations of the terms.

[0125] S150: Summarize the identified favorable factors, unfavorable factors, and handling recommendations to generate an analysis report.

[0126] Based on the analysis results, a detailed analysis report will be generated. The report will list the advantages and disadvantages of the contract in detail, and provide specific modification suggestions and action directions. The report will be presented in a user-friendly format (such as PDF or online browser format), including a clause summary, risk assessment, strengths analysis, and specific recommended measures.

[0127] The intelligent contract review method of this invention can provide users with comprehensive and in-depth contract review services, help users identify and mitigate contract risks, optimize contract structure, and thus improve the legal security and execution efficiency of contracts.

[0128] Example 2

[0129] This application provides a method for analyzing the strengths and weaknesses of contract text using a large model, including several functions such as document submission, selection of review perspective, extraction of document content, identification and analysis of content strengths and weaknesses, evaluation of content strengths and weaknesses, response and trade-offs for the evaluation, document content correction, and result summarization.

[0130] Step 1: Obtain the contract text to be reviewed via the chat dialog box;

[0131] Step 2: Obtain user input requirements through the chat dialog box, and provide the input requirements to the review model.

[0132] For example: Please analyze the contract history from the perspective of Party A, how to ensure that the contract terms fully protect your right to receive payment and avoid payment delays. Subsequent steps will be analyzed based on the current input requirements.

[0133] Step 3: Review the large model. By comprehensively using lexical, grammatical, semantic, and discourse analysis techniques, the text content is systematically processed to extract key information and clauses.

[0134] Specifically, this step first involves word segmentation, identifying each word in the text, and providing special understanding and marking for proper nouns, legal terms, and specific expressions. For example, it identifies contractual parties such as "Party A" and "Party B".

[0135] By analyzing the syntactic structure of the text, the grammatical relationships between words can be determined, such as which party's rights or obligations are followed by verbs like "enjoy" or "bear".

[0136] A deep understanding of the precise meaning of words and phrases in specific contexts is essential. When dealing with legal concepts such as rights and obligations and liability for breach of contract, it is crucial to understand and distinguish their subtle differences, such as recognizing expressions with specific legal meanings, like "shall perform," "have the right to claim," and "violate the agreement."

[0137] Considering the integrity and coherence of the contract text, the model needs to understand the interaction and constraint relationships between different clauses based on the context. For example, by identifying conditional sentences such as "If there is a breach of contract, then...", the model can find clauses related to liability for breach of contract.

[0138] Based on the above analysis results, the model will selectively extract key information such as the rights and obligations of Party A and Party B, liability for breach of contract, and dispute resolution methods (e.g., arbitration, litigation, or other agreed methods).

[0139] For example, the requirements in step 2 are: the model performs text segmentation to identify key entities and legal terms such as "Party A", "Party B", "payment", "delay", and "right to receive payment", and determines the attribution of rights and obligations in the clauses through syntactic analysis, such as identifying "Party A has the right to receive payment" through verb-object structure; and uses semantic understanding technology to distinguish the time constraint condition in "payment should be made within 30 days after receiving the goods".

[0140] Step 4: Review the large model to conduct a strengths and weaknesses analysis of the contract, recommend maintaining the favorable factors, and make recommendations for the unfavorable factors.

[0141] Specifically, in this step of conducting the advantages and disadvantages analysis, the review model will combine laws and regulations, historical cases, and specific clauses to comprehensively analyze the strengths and weaknesses of the contract.

[0142] It is recommended to maintain and strengthen favorable factors, such as: clearly listing the rights and obligations of both parties; and clearly defining product quality, service standards, and acceptance standards.

[0143] Suggestions for modification are made to address unfavorable factors, such as: clearly defining the boundaries of responsibility for specific matters, refining the attribution of responsibility in the event of breach of contract, and identifying such vague expressions in contract terms, such as "within a reasonable time" or "appropriate compensation," which may lead to disputes over interpretation during execution. The model will suggest clarifying quantitative standards or specific time limits, such as "within 30 working days" or "compensation at 120% of the actual loss amount."

[0144] Step 5: After the above steps are completed, the review model will summarize the analysis results of the contract text and generate a detailed report, including favorable factors, unfavorable factors, reference basis, response plans and modification suggestions.

[0145] The technical effects brought about by the above technical solution are as follows:

[0146] First, it greatly reduces the training cost of traditional algorithm models, eliminating the need for manual processes such as disassembling, labeling, classifying, training, and optimizing contract samples, thus eliminating heavy repetitive work.

[0147] Secondly, this large model exhibits superior performance, has a larger number of parameters, and is able to capture more complex data patterns and regularities.

[0148] Most importantly, the large-scale model possesses autonomous learning and optimization capabilities. Based on pre-trained data, it simultaneously stores the contract text as sample data in the base dataset during each contract text analysis. At the same time, it regularly updates legal documents and other materials, incorporating new laws and regulations into the base dataset. This continuous self-iteration and evolution during data processing, combined with user feedback and model output in practical applications, allows for ongoing model iteration and optimization of parameters and review rules based on actual conditions. Even when dealing with a small number of contract samples, it can make accurate answers and inferences based on its learned knowledge, ensuring the review model maintains high accuracy and effectiveness over the long term.

[0149] Example 3

[0150] like Figure 4 As shown, this embodiment provides a contract intelligent review system based on a large model, which includes:

[0151] The contract text acquisition module is used to acquire the contract texts that need to be reviewed;

[0152] The review requirements acquisition module is used to acquire the input review requirements;

[0153] The contract text processing module is used to process the contract text by comprehensively applying lexical analysis, syntactic analysis, semantic analysis and discourse analysis according to the review requirements, so as to extract key information and contract terms;

[0154] The contract advantages and disadvantages analysis module is used to perform an advantages and disadvantages analysis on the contract text based on the extracted key information and contract terms, and to identify favorable factors, unfavorable factors and handling suggestions.

[0155] The analysis report generation module is used to summarize the identified favorable factors, unfavorable factors, and handling suggestions to generate an analysis report.

[0156] In some embodiments, the contract text processing module specifically includes:

[0157] The lexical analysis submodule is used to identify each word in the contract text based on the word segmentation algorithm during lexical analysis, and to mark proper nouns, legal terms and specific expressions.

[0158] The syntax analysis submodule is used to parse the syntactic structure of the contract text based on the syntax parsing algorithm and determine the syntactic relationships between words during syntax analysis.

[0159] The semantic analysis submodule is used to determine the precise meaning of words and phrases in the contract text in the legal context based on semantic understanding algorithms when performing semantic analysis.

[0160] The document analysis submodule is used to determine the interaction and constraint relationships between different contract clauses based on the context of the contract text during document analysis, using a document analysis algorithm.

[0161] In some embodiments, the word segmentation algorithm includes: Jieba word segmentation algorithm, Tsinghua University Chinese word segmentation algorithm, or Harbin Institute of Technology Language Technology Platform word segmentation algorithm;

[0162] The syntactic parsing algorithms mentioned include: the Stanford parser or the Harbin Institute of Technology Language Technology Platform syntactic parser;

[0163] The semantic understanding algorithms include: the BERT model based on bidirectional encoder representation or the GPT model of generative pre-trained transformer;

[0164] The aforementioned text analysis algorithms include: TextRank or Latent Dirichlet Allocation (LDA).

[0165] In some embodiments, the contract advantages and disadvantages analysis module specifically includes:

[0166] The Contract Advantages and Disadvantages Determination Submodule is used to combine extracted key information, laws and regulations, historical cases, and contract clauses, and based on natural language processing algorithms and legal knowledge graphs, to determine the favorable and unfavorable factors of the contract text.

[0167] The favorable factors suggestion generation submodule is used to list the rights and obligations of the contracting parties based on the information extraction algorithm, and to clarify the product quality, service standards and acceptance standards for the aforementioned favorable factors.

[0168] The "Adverse Factors Modification Suggestion" submodule is used to propose modification suggestions for the aforementioned adverse factors, based on extracted key information and contract terms.

[0169] In some embodiments, the adverse factor modification suggestion submodule specifically includes:

[0170] The liability allocation suggestion unit is used to identify liability clauses in contract terms through rule engine algorithms and provide specific liability allocation suggestions in conjunction with legal knowledge graphs;

[0171] The fuzzy term processing unit is used to detect fuzzy terms in contracts through fuzzy matching algorithms and to provide clear quantitative standards or specific time limit suggestions using semantic analysis algorithms.

[0172] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to 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. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0173] Example 4

[0174] This invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements any of the above-described intelligent contract review methods based on a large model.

[0175] If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. Of course, there are other types of readable storage media, such as quantum memories, graphene memories, etc. It should be noted that the content contained in the computer-readable medium may be appropriately added to or subtracted from the content as required by the legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium may not include electrical carrier signals and telecommunication signals.

[0176] Example 5

[0177] The present invention also provides an electronic device. The electronic device according to embodiments of the present invention includes: one or more processors; and a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the large-model-based intelligent contract review method provided by the present invention.

[0178] The following is for reference. Figure 5 It shows a schematic diagram of the structure of a computer system 800 suitable for implementing an electronic device according to embodiments of the present invention. Figure 5 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention.

[0179] like Figure 5As shown, the computer system 500 includes a central processing unit (CPU) 501, which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) 502 or programs loaded from storage section 508 into random access memory (RAM) 503. The RAM 503 also stores various programs and data required for the operation of the computer system 500. The CPU 501, ROM 502, and RAM 503 are interconnected via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0180] The following components are connected to I / O interface 505: an input section 506 including a keyboard, mouse, etc.; an output section 507 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 508 including a hard disk, etc.; and a communication section 509 including a network interface card such as a LAN card, modem, etc. The communication section 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to I / O interface 505 as needed. A removable medium 511, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 510 as needed so that computer programs read from it can be installed into storage section 508 as needed.

[0181] In particular, according to the embodiments disclosed in this invention, the processes described in the above main step diagrams can be implemented as computer software programs. For example, embodiments of this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the main step diagrams. In the above embodiments, the computer program can be downloaded and installed from a network via communication section 509, and / or installed from removable medium 511. When the computer program is executed by central processing unit 501, it performs the functions defined in the system of this invention.

[0182] It should be noted that the computer-readable medium shown in this invention can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium may be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0183] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0184] The units described in the embodiments of the present invention can be implemented in software or hardware. The described units can also be located in a processor; for example, a processor can be described as including a front-end response unit, a receiving unit, and a request unit. The names of these units do not necessarily limit the specific unit itself.

[0185] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A contract intelligent review method based on a large model, characterized in that, Includes the following steps: Obtain the contract text that needs to be reviewed; Obtain the review requirements for the input; Based on the aforementioned review requirements, the contract text is processed using a combination of lexical analysis, syntactic analysis, semantic analysis, and discourse analysis to extract key information and contract terms. Based on the extracted key information and contract terms, the contract text is analyzed for advantages and disadvantages using a large model to identify favorable factors, unfavorable factors, and handling suggestions. The identified favorable factors, unfavorable factors, and handling recommendations are summarized to generate an analysis report; The aforementioned method, based on extracted key information and contract terms, performs a strengths and weaknesses analysis of the contract text using a large model to identify favorable factors, unfavorable factors, and proposed solutions. Specifically, this includes: By combining extracted key information, laws and regulations, historical cases, and contract terms, and based on natural language processing algorithms and legal knowledge graphs, the favorable and unfavorable factors of the contract text are determined. Based on the aforementioned favorable factors, the rights and obligations of the contracting parties are listed using an information extraction algorithm, and the product quality, service standards, and acceptance standards are clearly defined. Based on the aforementioned adverse factors, and in conjunction with the extracted key information and contract terms, modification suggestions are proposed. The proposed modifications, based on the extracted key information and contract terms, address unfavorable factors and include: The algorithm uses a rule engine to identify liability clauses in contracts and combines them with a legal knowledge graph to provide specific suggestions on liability allocation. The fuzzy matching algorithm is used to detect ambiguous terms in contracts, and the semantic analysis algorithm is used to provide clear quantitative standards or specific time limits.

2. The method according to claim 1, characterized in that, The aforementioned method, based on the review requirements, comprehensively utilizes lexical analysis, syntactic analysis, semantic analysis, and discourse analysis to process the contract text in order to extract key information and contract clauses, specifically including: During lexical analysis, each word in the contract text is identified based on a word segmentation algorithm, and proper nouns, legal terms, and specific expressions are marked. During the syntactic analysis, the syntactic structure of the contract text is analyzed based on the syntactic parsing algorithm to determine the syntactic relationships between words; During semantic analysis, semantic understanding algorithms are used to determine the precise meaning of words and phrases in the contract text within the legal context. When performing discourse analysis, the discourse analysis algorithm determines the interaction and constraint relationships between different contract clauses based on the context of the contract text.

3. The method according to claim 2, characterized in that, The word segmentation algorithms include: Jieba word segmentation algorithm, Tsinghua University Chinese word segmentation algorithm, or Harbin Institute of Technology Language Technology Platform word segmentation algorithm; The syntactic parsing algorithms mentioned include: the Stanford parser or the Harbin Institute of Technology Language Technology Platform syntactic parser; The semantic understanding algorithms include: the BERT model based on bidirectional encoder representation or the GPT model of generative pre-trained transformer; The aforementioned text analysis algorithms include: TextRank or Latent Dirichlet Allocation (LDA).

4. A contract intelligent review system based on a large model, characterized in that, include: The contract text acquisition module is used to acquire the contract texts that need to be reviewed; The review requirements acquisition module is used to acquire the input review requirements; The contract text processing module is used to process the contract text by comprehensively applying lexical analysis, syntactic analysis, semantic analysis and discourse analysis according to the review requirements, so as to extract key information and contract terms; The contract advantages and disadvantages analysis module is used to perform an advantages and disadvantages analysis on the contract text based on the extracted key information and contract terms, and to identify favorable factors, unfavorable factors and handling suggestions based on a large model. The analysis report generation module is used to summarize the identified favorable factors, unfavorable factors, and handling suggestions to generate an analysis report; The contract advantages and disadvantages analysis module specifically includes: The Contract Advantages and Disadvantages Determination Submodule is used to combine extracted key information, laws and regulations, historical cases, and contract clauses, and based on natural language processing algorithms and legal knowledge graphs, to determine the favorable and unfavorable factors of the contract text. The favorable factors suggestion generation submodule is used to list the rights and obligations of the contracting parties based on the information extraction algorithm, and to clarify the product quality, service standards and acceptance standards for the aforementioned favorable factors. The "Adverse Factors Modification Suggestion" submodule is used to propose modification suggestions for the aforementioned adverse factors, based on extracted key information and contract terms. The submodule for suggesting modifications to adverse factors specifically includes: The liability allocation suggestion unit is used to identify liability clauses in contract terms through rule engine algorithms and provide specific liability allocation suggestions in conjunction with legal knowledge graphs; The fuzzy term processing unit is used to detect fuzzy terms in contracts through fuzzy matching algorithms and to provide clear quantitative standards or specific time limit suggestions using semantic analysis algorithms.

5. The system according to claim 4, characterized in that, The contract text processing module specifically includes: The lexical analysis submodule is used to identify each word in the contract text based on the word segmentation algorithm during lexical analysis, and to mark proper nouns, legal terms and specific expressions. The syntax analysis submodule is used to parse the syntactic structure of the contract text based on the syntax parsing algorithm and determine the syntactic relationships between words during syntax analysis. The semantic analysis submodule is used to determine the precise meaning of words and phrases in the contract text in the legal context based on semantic understanding algorithms when performing semantic analysis. The document analysis submodule is used to determine the interaction and constraint relationships between different contract clauses based on the context of the contract text during document analysis, using a document analysis algorithm.

6. The system according to claim 5, characterized in that, The word segmentation algorithms include: Jieba word segmentation algorithm, Tsinghua University Chinese word segmentation algorithm, or Harbin Institute of Technology Language Technology Platform word segmentation algorithm; The syntactic parsing algorithms mentioned include: the Stanford parser or the Harbin Institute of Technology Language Technology Platform syntactic parser; The semantic understanding algorithms include: the BERT model based on bidirectional encoder representation or the GPT model of generative pre-trained transformer; The aforementioned text analysis algorithms include: TextRank or Latent Dirichlet Allocation (LDA).

Images