Contract approval method and device, equipment and medium

Abstract

The invention provides a contract approval method and device, equipment and a medium, and the method comprises the steps: analyzing a to-be-approved contract text, and extracting entities in the contract text and a semantic relation between the entities; instantiating the extracted entities and semantic relationships into dynamic sub-graphs; associating the dynamic sub-graph with a pre-constructed static knowledge graph to obtain an association result, the static knowledge graph storing laws and regulations and enterprise approval rules and defining entity types and semantic relationship types; on the basis of the association result, performing dual verification on terms in the contract text to obtain a dual verification result, the dual verification including semantic consistency verification based on a static knowledge graph and logic compliance verification based on an enterprise business process rule; and generating a natural language approval opinion with a legal basis and / or an enterprise rule basis according to a double verification result.

Description

Technical Field

[0001] This application relates to the field of big data technology, specifically to a contract approval method, apparatus, equipment, and medium. Background Technology

[0002] Traditional contract approval processes heavily rely on manual review by legal and business personnel. This model suffers from significant bottlenecks in efficiency, standards, and risk control. Reviewers must meticulously examine unstructured contract texts and compare them against a vast amount of laws, regulations, and internal company policies, resulting in approval cycles of several working days for a single contract and severely hindering business efficiency. More problematic is the difficulty in ensuring consistent standards and comprehensive judgment through manual review. Different reviewers may have misunderstandings of the same clause that can differ by as much as 40%. Furthermore, manual verification suffers from a high rate of omissions regarding complex or hidden legal risks, exposing companies to potential legal and operational risks.

[0003] To address these challenges, existing technologies have proposed some automation solutions, but all have inherent limitations. While pure rule-based engines can handle simple logic such as "amounts > 1 million require higher-level approval," they cannot understand the deeper semantics of contract terms and struggle to handle complex contexts and intent interpretations in legal texts. General-purpose natural language processing models, although possessing basic text understanding capabilities, lack the support of a professional legal knowledge system, resulting in insufficient accuracy and professionalism in their outputs, making them unsuitable for direct application in serious legal compliance scenarios. Therefore, the industry urgently needs a new solution that deeply integrates domain expertise with advanced artificial intelligence technology to achieve a synergistic improvement in the efficiency, quality, and reliability of contract approval. Summary of the Invention

[0004] In view of the above problems, this application provides a contract approval method, apparatus, equipment and medium that solves at least one of the above problems.

[0005] According to a first aspect of this application, a contract approval method is provided, comprising: parsing a contract text to be approved and extracting entities and semantic relationships between entities from the contract text; instantiating the extracted entities and semantic relationships into a dynamic subgraph; associating the dynamic subgraph with a pre-constructed static knowledge graph to obtain an association result, wherein the static knowledge graph stores laws and regulations, enterprise approval rules, and defines entity types and semantic relationship types; performing dual verification on the clauses in the contract text based on the association result to obtain a dual verification result, wherein the dual verification includes semantic consistency verification based on the static knowledge graph and logical compliance verification based on enterprise business process rules; and generating a natural language approval opinion with legal basis and / or enterprise rule basis based on the dual verification result.

[0006] In some embodiments, parsing the contract text to be approved and extracting entities and semantic relationships between entities in the contract text includes: using the BERT-CRF model to perform named entity recognition on the contract text to extract contract entities; and extracting semantic relationships in the form of triples, including subject, behavior, and object, from the contract text based on dependency syntax and a legal domain dictionary.

[0007] In some embodiments, associating the dynamic subgraph with a pre-built static knowledge graph to obtain an association result includes: semantically matching the contract entities in the dynamic subgraph with the nodes in the static knowledge graph through entity links; and establishing semantic association edges between the successfully matched dynamic subgraph nodes and static knowledge graph nodes to mount the dynamic subgraph to the static knowledge graph.

[0008] In some embodiments, generating a natural language approval opinion with legal basis includes: calling a predefined text template containing legal basis variables; filling the corresponding variables of the text template with the result of the double verification, wherein the result of the double verification includes risk type, relevant contract terms, relevant legal provisions or corporate rules, and suggested actions; and selecting the corresponding text template for output based on the risk level of the result of the double verification.

[0009] In some embodiments, performing semantic consistency verification on the clauses in the contract text based on the association result, using the static knowledge graph, includes: calculating the annualized ratio corresponding to the daily penalty rate in the contract; querying the upper limit of the penalty rate stipulated in relevant legal provisions from the static knowledge graph; comparing whether the annualized ratio exceeds the upper limit to obtain a comparison result; and generating a semantic consistency verification result based on the comparison result.

[0010] In some embodiments, performing semantic consistency verification on the clauses in the contract text based on the static knowledge graph based on the association results includes: locating legal and regulatory nodes related to the contract entity by traversing the association path between the dynamic subgraph and the static knowledge graph; calculating the semantic matching degree or conflict relationship between the content of the contract clauses and the content of the legal and regulatory nodes; and determining whether the contract clauses are legal and compliant based on the matching degree or conflict relationship.

[0011] In some embodiments, performing logical compliance verification based on enterprise business process rules on the terms in the contract text based on the association results includes: parsing the key business elements of the contract from the association results between the dynamic subgraph and the static knowledge graph; matching the key business elements with predefined approval rule conditions of the enterprise; and determining whether the contract triggers the corresponding business process or violates the enterprise's internal system based on the matching results.

[0012] According to a second aspect of this application, a contract approval device is provided, comprising: an extraction module for parsing a contract text to be approved and extracting entities and semantic relationships between entities from the contract text; an instantiation module for instantiating the extracted entities and semantic relationships into a dynamic subgraph; an association module for associating the dynamic subgraph with a pre-constructed static knowledge graph to obtain an association result, wherein the static knowledge graph stores laws and regulations, enterprise approval rules, and defines entity types and semantic relationship types; a verification module for performing dual verification on the clauses in the contract text based on the association result to obtain a dual verification result, wherein the dual verification includes semantic consistency verification based on the static knowledge graph and logical compliance verification based on enterprise business process rules; and a generation module for generating natural language approval opinions with legal basis and / or enterprise rule basis based on the dual verification result.

[0013] According to a third aspect of this application, an electronic device is provided, comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the method according to the first aspect.

[0014] According to a fourth aspect of this application, a computer-readable storage medium is provided having a computer program or instructions stored thereon, which, when executed by a processor, implement the steps of the method according to the first aspect. Attached Figure Description

[0015] The above-mentioned contents, other objects, features and advantages of this application will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0016] Figure 1 A flowchart illustrating a contract approval method according to an embodiment of this application is shown schematically;

[0017] Figure 2 This schematic diagram illustrates the structure of a contract approval device according to an embodiment of this application;

[0018] Figure 3 A schematic diagram of an electronic device according to an embodiment of this application is shown. Detailed Implementation

[0019] The embodiments of this application will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of this application. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of this application for ease of explanation. However, it will be apparent that one or more embodiments may be implemented without these specific details. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concepts of this application.

[0020] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0021] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0022] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0023] In the technical solution disclosed herein, the user information (including but not limited to user personal information, user image information, user transaction record information, user device information such as location information) and data (including but not limited to data used for analysis, stored data, and displayed data) involved are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of related data all comply with relevant laws, regulations, and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entry points for users to choose to authorize or refuse.

[0024] In scenarios involving automated decision-making using personal information, the methods, devices, and systems provided in this disclosure all offer users corresponding entry points for choosing to agree to or reject the automated decision-making results. If the user chooses to reject, the process proceeds to the expert decision-making stage. Here, "automated decision-making" refers to the activity of automatically analyzing and evaluating an individual's behavioral habits, interests, or economic, health, and credit status through computer programs, and then making a decision. Here, "expert decision-making" refers to the activity of making decisions by personnel who specialize in a particular field, possess specialized experience, knowledge, and skills, and have reached a certain level of professional expertise.

[0025] Figure 1 A flowchart illustrating a contract approval method according to an embodiment of this application is shown schematically.

[0026] like Figure 1 As shown, the contract approval method includes operations S110-S150.

[0027] In operation S110, the contract text to be approved is parsed, and the entities in the contract text and the semantic relationships between the entities are extracted.

[0028] According to embodiments of the present invention, an entity can refer to a key information element with specific legal or business significance identified from the contract text. Specifically, these entities may include the contracting parties, contract amount, payment terms, liability clauses for breach of contract, descriptions of goods or services, etc.

[0029] According to embodiments of the present invention, semantic relations can refer to relations that connect the aforementioned entities and express specific legal acts or logical connections. Specifically, techniques such as dependency parsing can be used to extract "subject-behavior-object" triplet relations such as "(Party A) - [Payment] -> (Payment for Goods)" and "(Party B) - [Assume] -> (Liability for Breach of Contract)".

[0030] In operation S120, the extracted entities and semantic relations are instantiated into dynamic subgraphs.

[0031] According to embodiments of the present invention, instantiation refers to creating corresponding data objects and relationships in computer memory or graph database.

[0032] According to embodiments of the present invention, a dynamic subgraph can be a structured graph data model temporarily generated to represent the currently approved contract. The dynamic aspect is that it is temporarily constructed for a single contract. The subgraph can also be associated with a larger main graph. This dynamic subgraph consists of nodes and edges, where nodes represent the extracted entities and edges represent the extracted "semantic relationships."

[0033] In operation S130, the dynamic subgraph is associated with the pre-built static knowledge graph to obtain the association result.

[0034] According to embodiments of the present invention, a static knowledge graph stores laws and regulations, enterprise approval rules, and defines entity types and semantic relationship types. The static knowledge graph can be a pre-constructed, large-scale semantic network storing general domain knowledge. In embodiments of the present invention, it can store textual knowledge from legal databases (such as the Civil Code and Contract Law) and business rule knowledge from internal enterprise regulations (such as the Financial Approval Manual and Risk Control Standards). Nodes in the graph can be defined as entity types such as "legal provisions," "compliance standards," and "approval roles," while edges are defined as semantic relationship types such as "reference," "violation," and "constraint."

[0035] According to an embodiment of the present invention, entity linking technology can be used to semantically match contract entities (such as "20% penalty for breach of contract") in a dynamic subgraph with corresponding nodes (such as nodes representing Article 585 of the Civil Code) in a static knowledge graph, and temporary association edges can be established between successfully matched nodes to form an association, so that the contract content can be examined in the context of knowledge.

[0036] In operation S140, based on the association results, double verification is performed on the clauses in the contract text to obtain the results of the double verification.

[0037] According to embodiments of the present invention, the dual verification includes semantic consistency verification based on static knowledge graphs and logical compliance verification based on enterprise business process rules.

[0038] According to embodiments of the present invention, semantic consistency verification based on static knowledge graphs refers to using the semantic reasoning capabilities of the graph to verify whether the substantive content of contract terms conflicts with the spirit of the law or specific provisions. For example, by associating legal nodes related to "liquidation penalty," the proportion of liquidated damages stipulated in the contract can be calculated to see if it exceeds the legally prescribed upper limit.

[0039] According to embodiments of the present invention, logical compliance verification based on enterprise business process rules refers to checking whether a contract triggers a specific approval process or violates internal regulations based on the enterprise's predefined business logic. For example, it determines whether the contract amount exceeds the current approver's authority, thereby automatically prompting that a higher level of approval is required.

[0040] In operation S150, based on the results of the dual verification, a natural language approval opinion with legal basis and / or corporate rules is generated.

[0041] According to an embodiment of the present invention, the verification results (such as "violation of Article XX of the XX Law" or "amount exceeding the limit requires XX approval") can be used as structured input and filled into a predefined text template containing legal basis and / or corporate rules. This will automatically combine the results into a complete and interpretable natural language approval opinion, such as: "The liquidated damages ratio stipulated in Article 3 of the contract may exceed the reasonable range stipulated in Article 585 of the Civil Code. Legal review is recommended. At the same time, since the contract amount exceeds RMB 1 million, a board resolution is required."

[0042] According to embodiments of the present invention, unstructured contract text is transformed into structured dynamic subgraphs using natural language processing technology, and semantically associated with a static knowledge graph containing domain knowledge. Based on this, a dual verification mechanism is executed, which verifies the semantic consistency of the knowledge graph and the compliance of enterprise rule logic. Finally, a legally based approval opinion is generated, realizing the transformation of contract approval from human experience judgment to automated intelligent decision-making, significantly improving approval efficiency, ensuring the objectivity and consistency of approval standards, and effectively reducing legal risks caused by human oversight.

[0043] In some embodiments of the present invention, the BERT-CRF model can be used to perform named entity recognition on the contract text in order to extract contract entities; based on dependency syntax and legal domain dictionaries, semantic relations in the form of triplets including subject, behavior and object can be extracted from the contract text.

[0044] The BERT-CRF model can be a specific hybrid neural network model used to perform named entity recognition tasks. The BERT (Bidirectional Encoder Representation from Transformer) model acts as a feature extractor, responsible for deeply understanding the bidirectional semantic information of each character / word in the contract text within its context, thus better addressing common issues like polysemy and long-range dependencies in Chinese legal contracts. The CRF (Conditional Random Field) model, acting as a label decoder, sits atop the BERT model and is responsible for considering the transition rules and constraints between labels during sequence labeling, ensuring the grammatical and business logic rationality of the output entity label sequences (such as "B-Party, I-Party, O..."). This combination can improve the accuracy of entity recognition in the vertical domain of legal contracts.

[0045] A contractual entity can refer to a named unit of information with specific legal or commercial significance identified from the contract text. Contractual entities can include the following types: contracting parties,

[0046] The amount and payment entity, time and deadline entity, liability and default entity, and subject matter entity, etc.

[0047] Dependency parsing can construct a syntactic dependency tree by analyzing the grammatical dependency relationships between words in a sentence, such as modification, subject-verb, and verb-object relationships. For example, in the sentence "Party A should pay Party B for the goods," dependency parsing can determine that "pay" is the core predicate verb, "Party A" is its subject, and "goods payment" is its object.

[0048] Legal dictionaries can pre-include core verbs and patterns representing key legal acts or relationships in the field of legal contracts, such as: {"payment", "delivery", "compensation", "guarantee", "authorization", "breach of contract"}, etc. During relation extraction, attention can be paid to the syntactic structures related to these dictionary terms, thereby filtering out a large amount of irrelevant information and improving the accuracy and efficiency of relation extraction.

[0049] The subject can refer to the initiator of the legal act, the act can refer to the core legal action jointly determined by the domain dictionary and syntactic analysis, and the object can refer to the recipient of the legal act. Ultimately, the system transforms contract terms into a series of standard triples such as (Party A, payment, payment for goods), (Party B, delivery, goods), and (either party, breach, contract). These triples provide a direct data foundation for subsequently constructing precise dynamic subgraphs.

[0050] In some embodiments of the present invention, the contract entities in the dynamic subgraph can be semantically matched with the nodes in the static knowledge graph through entity links; semantic association edges are established between the successfully matched dynamic subgraph nodes and the static knowledge graph nodes to mount the dynamic subgraph to the static knowledge graph.

[0051] For example, firstly, the names and attribute texts of contract entities in the dynamic subgraph (such as "overdue penalty") and candidate nodes in the static knowledge graph (such as nodes for "penalty", "late payment fee", "compensation") are converted into vector representations. Then, vector similarity is calculated in a high-dimensional semantic space. If the similarity score exceeds a preset threshold, a successful match is determined.

[0052] Furthermore, the matching process can also utilize contextual information, namely, by observing other nodes and edges connected to the entity in the dynamic subgraph, to assist in disambiguation. For example, when the contract entity "Gold" appears in a payment context, it is more likely to be linked to the "Amount" node; while when it appears in a default context, it is more likely to be linked to the "Penalty" node.

[0053] Semantic edges can be specific data structures used to connect dynamic data and static knowledge. They are temporary, directed or undirected connections created in a graph database to logically represent a semantic relationship between nodes in a dynamic subgraph and nodes in a static knowledge graph.

[0054] In some embodiments of the present invention, a predefined text template containing legal basis variables can be invoked; the results of the double verification are filled into the corresponding variables of the text template. The results of the double verification include risk type, relevant contract terms, relevant legal provisions or corporate rules, and suggested operations; the corresponding text template is selected for output based on the risk level of the double verification results.

[0055] A text template can be a pre-defined text frame containing fixed text and specific placeholder variables. The fixed text is a standard expression conforming to legal document norms, while the placeholder variables are used to insert specific, dynamic information derived from double-checking, such as {legal provisions}, {contract terms}, {risk levels}, etc.

[0056] The result of the double verification can be the structured data input required to fill the template, which may include risk type, relevant contract terms, relevant legal provisions or company rules, suggested actions, etc.

[0057] In some embodiments of the present invention, the annualized ratio corresponding to the daily liquidated damages ratio in the contract can be calculated; the upper limit of the liquidated damages ratio stipulated in relevant legal provisions can be queried from the static knowledge graph; whether the annualized ratio exceeds the upper limit can be compared to obtain the comparison result; and a semantic consistency verification result can be generated based on the comparison result.

[0058] In some embodiments of the present invention, the legal and regulatory nodes related to the contract entity can be located by traversing the association paths between the dynamic subgraph and the static knowledge graph; the semantic matching degree or conflict relationship between the contract terms and the legal and regulatory nodes can be calculated; and the legality and compliance of the contract terms can be determined based on the matching degree or conflict relationship.

[0059] Legal and regulatory nodes refer to standardized data entities in a static knowledge graph that represent specific legal provisions, administrative regulations, judicial interpretations, etc. Each node contains attributes such as "law name," "provision number," "provision content," and "applicable circumstances."

[0060] For example, using graph query languages ​​(such as Cypher or Gremlin), starting from a contract entity node in a dynamic subgraph, one can perform single-hop or multi-hop graph traversal along established "semantic association edges" (such as REFERS_TO) and predefined relation edges in a static knowledge graph (such as LEGAL_BASIS) to discover all related legal and regulatory nodes.

[0061] Calculating semantic matching or conflict relationships can include numerical logic calculations and semantic similarity calculations. Numerical logic calculations involve comparing quantitative indicators extracted from the contract with quantitative standards in legal provisions for clauses involving numerical values, proportions, or timeframes. For example, the system might calculate whether the contractually stipulated "daily penalty of 0.1%" results in an "annualized interest rate of 36.5%" and compare this value with the legal provision that "it shall not exceed 30% of the losses incurred," thereby identifying a "conflict relationship."

[0062] Semantic similarity calculation can be used to assess the matching degree of contractual terms and legal provisions that involve qualitative descriptions by converting the content of the contractual terms and legal provisions into semantic vectors and calculating the cosine similarity between the vectors. For example, it can be used to determine whether the definition of "force majeure" in the contract is consistent with the definition in the law.

[0063] In some embodiments of the present invention, key business elements of a contract can be parsed from the association results of dynamic subgraphs and static knowledge graphs; the key business elements can be matched with predefined approval rules and conditions of the enterprise; and based on the matching results, it can be determined whether the contract triggers the corresponding business process or violates the enterprise's internal system.

[0064] Key business elements can be structured data that is directly extracted or indirectly derived from the entity node attributes of dynamic subgraphs and their association with static graphs. Key business elements can include monetary elements, subject elements, time elements, and clause elements, etc.

[0065] For example, the parsed key business elements can be used as input variables and substituted into the rule conditions for logical operations. If the value of a business element satisfies a condition of a rule, that rule is triggered.

[0066] Based on the above contract approval method, this application also provides a contract approval device. The following will combine... Figure 2 The device is described in detail.

[0067] Figure 2 A schematic block diagram of a contract approval device according to an embodiment of this application is shown.

[0068] like Figure 2 As shown, the contract approval device in this embodiment includes an extraction module 310, an instantiation module 320, an association module 330, a verification module 340, and a generation module 350.

[0069] The extraction module 310 is used to parse the contract text to be approved and extract the entities in the contract text and the semantic relationships between the entities. In one embodiment, the extraction module 310 can be used to perform the operation S110 described above, which will not be repeated here.

[0070] The instantiation module 320 is used to instantiate the extracted entities and semantic relationships into a dynamic subgraph; in one embodiment, the instantiation module 320 can be used to perform the operation S120 described above, which will not be repeated here.

[0071] The association module 330 is used to associate the dynamic subgraph with a pre-built static knowledge graph to obtain the association result. The static knowledge graph stores laws and regulations, enterprise approval rules, and defines entity types and semantic relationship types. In one embodiment, the association module 330 can be used to perform the operation S130 described above, which will not be repeated here.

[0072] The verification module 340 performs dual verification on the clauses in the contract text based on the association results, obtaining the results of the dual verification. The dual verification includes semantic consistency verification based on a static knowledge graph and logical compliance verification based on enterprise business process rules. In one embodiment, the verification module 340 can be used to perform the operation S140 described above, which will not be repeated here.

[0073] The generation module 350 is used to generate natural language approval opinions with legal basis and / or corporate rules basis based on the results of the double verification. In one embodiment, the generation module 350 can be used to perform the operation S150 described above, which will not be repeated here.

[0074] In some embodiments, parsing the contract text to be approved and extracting entities and semantic relationships between entities in the contract text includes: using the BERT-CRF model to perform named entity recognition on the contract text to extract contract entities; and extracting semantic relationships in the form of triplets, including subjects, behaviors, and objects, from the contract text based on dependency syntax and legal domain dictionaries.

[0075] In some embodiments, associating a dynamic subgraph with a pre-built static knowledge graph to obtain association results includes: semantically matching contract entities in the dynamic subgraph with nodes in the static knowledge graph through entity links; and establishing semantic association edges between successfully matched dynamic subgraph nodes and static knowledge graph nodes to mount the dynamic subgraph to the static knowledge graph.

[0076] In some embodiments, generating a natural language approval opinion with legal basis includes: calling a predefined text template containing legal basis variables; filling the corresponding variables in the text template with the results of double verification, the results of double verification including risk type, relevant contract terms, relevant legal provisions or corporate rules, and suggested actions; and selecting the corresponding text template for output based on the risk level of the double verification results.

[0077] In some embodiments, the semantic consistency verification of the clauses in the contract text based on the association results includes: calculating the annualized ratio corresponding to the daily liquidated damages ratio in the contract; querying the upper limit of the liquidated damages ratio stipulated in the relevant legal provisions from the static knowledge graph; comparing whether the annualized ratio exceeds the upper limit to obtain the comparison result; and generating the semantic consistency verification result based on the comparison result.

[0078] In some embodiments, performing semantic consistency verification on the clauses in the contract text based on the association results includes: locating legal and regulatory nodes related to the contract entity by traversing the association paths between the dynamic subgraph and the static knowledge graph; calculating the semantic matching degree or conflict relationship between the content of the contract clauses and the content of the legal and regulatory nodes; and determining whether the contract clauses are legal and compliant based on the matching degree or conflict relationship.

[0079] In some embodiments, the logical compliance verification of the clauses in the contract text based on the enterprise business process rules, based on the association results, includes: parsing the key business elements of the contract from the association results of the dynamic subgraph and the static knowledge graph; matching the key business elements with the enterprise's predefined approval rule conditions; and determining whether the contract triggers the corresponding business process or violates the enterprise's internal system based on the matching results.

[0080] According to embodiments of this application, any multiple modules among the feature extraction module 310, mapping module 320, substitution module 330, fusion module 340, and generation module 350 can be merged into one module, or any one of these modules can be split into multiple modules. Alternatively, at least some of the functions of one or more of these modules can be combined with at least some of the functions of other modules and implemented in one module. According to embodiments of this application, at least one of the feature extraction module 310, mapping module 320, substitution module 330, fusion module 340, and generation module 350 can be at least partially implemented as hardware circuitry, such as a field-programmable gate array (FPGA), a programmable logic array (PLA), a system-on-a-chip, a system-on-a-substrate, a system-on-package, an application-specific integrated circuit (ASIC), or implemented in hardware or firmware by any other reasonable means of integrating or packaging the circuitry, or implemented in software, hardware, or firmware, or in any suitable combination of any of these three implementation methods. Alternatively, at least one of the feature extraction module 310, mapping module 320, substitution module 330, fusion module 340, and generation module 350 may be implemented at least partially as a computer program module, which can perform corresponding functions when the computer program module is run.

[0081] Figure 3A block diagram schematically illustrates an electronic device suitable for implementing a user profile construction method according to an embodiment of this application.

[0082] like Figure 3 As shown, an electronic device according to an embodiment of this application includes a processor 401, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 402 or a program loaded from a storage portion 408 into a random access memory (RAM) 403. The processor 401 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 401 may also include onboard memory for caching purposes. The processor 401 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of this application.

[0083] RAM 403 stores various programs and data required for the operation of the electronic device. Processor 401, ROM 402, and RAM 403 are interconnected via bus 404. Processor 401 executes various operations of the method flow according to embodiments of this application by executing programs in ROM 402 and / or RAM 403. It should be noted that programs may also be stored in one or more memories other than ROM 402 and RAM 403. Processor 401 may also execute various operations of the method flow according to embodiments of this application by executing programs stored in said one or more memories.

[0084] According to embodiments of this application, the electronic device may further include an input / output (I / O) interface 405, which is also connected to a bus 404. The electronic device may also include one or more of the following components connected to the input / output (I / O) interface 405: an input section 406 including a keyboard, mouse, etc.; an output section 407 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 408 including a hard disk, etc.; and a communication section 409 including a network interface card such as a LAN card, modem, etc. The communication section 409 performs communication processing via a network such as the Internet. A drive 410 is also connected to the input / output (I / O) interface 405 as needed. A removable medium 411, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 410 as needed so that computer programs read from it can be installed into the storage section 408 as needed.

[0085] This application also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to the embodiments of this application.

[0086] According to embodiments of this application, the computer-readable storage medium can be a non-volatile computer-readable storage medium, such as including but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, the 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. For example, according to embodiments of this application, the computer-readable storage medium may include ROM 402 and / or RAM 403 and / or one or more memories other than ROM 402 and RAM 403 described above.

[0087] Embodiments of this application also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowchart. When the computer program product is run on a computer system, the program code is used to enable the computer system to implement the contract approval method provided in the embodiments of this application.

[0088] When the computer program is executed by the processor 401, it performs the functions defined in the system / apparatus of this application embodiment. According to the embodiments of this application, the systems, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0089] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and downloaded and installed via communication section 409, and / or installed from removable medium 411. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0090] In such an embodiment, the computer program can be downloaded and installed from a network via communication section 409, and / or installed from removable medium 411. When the computer program is executed by processor 401, it performs the functions defined in the system of this application embodiment. According to embodiments of this application, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0091] According to embodiments of this application, program code for executing the computer programs provided in the embodiments of this application can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0092] 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 this application. 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.

[0093] Those skilled in the art will understand that the features described in the various embodiments of this application can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in this application. In particular, the features described in the various embodiments of this application can be combined and / or combined in various ways without departing from the spirit and teachings of this application. All such combinations and / or combinations fall within the scope of this application.

Claims

1. A contract approval method characterized by, The method comprises the following steps: parsing a to-be-approved contract text, extracting entities and semantic relationships between the entities in the contract text; instantiating the extracted entities and semantic relationships as dynamic subgraphs; associating the dynamic subgraphs with a pre-constructed static knowledge graph to obtain an association result, the static knowledge graph storing laws and regulations and enterprise approval rules and defining entity types and semantic relationship types; based on the association result, performing double verification on the clauses in the contract text to obtain a double verification result, the double verification including semantic consistency verification based on the static knowledge graph and logical compliance verification based on enterprise business process rules; generating a natural language approval opinion with legal basis and / or enterprise rule basis according to the double verification result.

2. The method of claim 1, wherein, The parsing of the to-be-approved contract text and the extraction of the entities and the semantic relationships between the entities in the contract text comprise: performing named entity recognition on the contract text using a BERT-CRF model to extract contract entities; extracting semantic relationships in the form of triples including subjects, actions and objects from the contract text based on dependency syntax and a legal domain dictionary.

3. The method of claim 1, wherein, The association of the dynamic subgraphs with the pre-constructed static knowledge graph to obtain an association result comprises: performing semantic matching between the contract entities in the dynamic subgraphs and the nodes in the static knowledge graph through entity linking; establishing semantic association edges between the matched dynamic subgraph nodes and static knowledge graph nodes to mount the dynamic subgraphs to the static knowledge graph.

4. The method of claim 1, wherein, The generation of a natural language approval opinion with legal basis comprises: calling a predefined text template containing legal basis variables; filling the double verification result into the corresponding variables of the text template, the double verification result including risk type, related contract clause, involved legal provisions or enterprise rules, and suggested operation; selecting a corresponding text template for output according to the risk level of the double verification result.

5. The method of claim 1, wherein, The semantic consistency verification based on the static knowledge graph performed on the clauses in the contract text based on the association result comprises: calculating the annualized ratio corresponding to the liquidated damages ratio in the contract; querying the upper limit of the liquidated damages ratio specified in the relevant legal provisions from the static knowledge graph; comparing whether the annualized ratio exceeds the upper limit to obtain a comparison result; generating a semantic consistency verification result according to the comparison result.

6. The method of claim 1, wherein, The semantic consistency verification based on the static knowledge graph performed on the clauses in the contract text based on the association result comprises: locating the legal regulation nodes related to the contract entities by traversing the association path between the dynamic subgraphs and the static knowledge graph; calculating the semantic matching degree or conflict relationship between the contract clause content and the legal regulation node content; judging whether the contract clause is legal and compliant based on the matching degree or conflict relationship.

7. The method of claim 1, wherein, The logical compliance verification based on enterprise business process rules performed on the clauses in the contract text based on the association result comprises: From the association result of the dynamic subgraph and the static knowledge graph, the key business elements of the contract are parsed out; The key business elements are matched with the pre-defined approval rule conditions of the enterprise; According to the matching result, it is judged whether the contract triggers the corresponding business process or violates the internal system of the enterprise.

8. A contract approval device characterized by comprising: Comprise: The extraction module is used to parse the to-be-approved contract text, extract the entities and semantic relationships between the entities in the contract text; The instantiation module is used to instantiate the extracted entities and semantic relationships into a dynamic subgraph; The association module is used to associate the dynamic subgraph with the pre-constructed static knowledge graph to obtain an association result, and the static knowledge graph stores laws and regulations, enterprise approval rules, and defines entity types and semantic relationship types; The verification module is used to perform double verification on the clauses in the contract text based on the association result, and obtain a double verification result, the double verification including semantic consistency verification based on the static knowledge graph and logical compliance verification based on the enterprise business process rules; The generation module is used to generate natural language approval opinions with legal basis and / or enterprise rule basis according to the double verification result.

9. An electronic device comprising: One or more processors; Memory for storing one or more computer programs, characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1-7.

10. A computer readable storage medium having stored thereon a computer program or instructions, characterized in that, The computer program or instructions are executed by the processor to implement the steps of the method according to any one of claims 1-7.

Images