Intelligent risk control compliance method of contract

Abstract

The invention discloses an intelligent risk control compliance method for a contract, and relates to the technical field of risk control compliance management, and the method comprises the steps: obtaining a to-be-processed contract, carrying out the structural analysis of the to-be-processed contract, and extracting the structural contract data corresponding to the to-be-processed contract; performing intelligent review on the structured contract data based on a preset compliance rule base, and generating risk nodes in the structured contract data; multi-source external data of a preset platform are collected, contract related data matched with the structured contract data are extracted from the multi-source external data, and the multi-source external data comprise judicial data, administrative data and public opinion data; and inputting the structured contract data, the risk node and the contract related data into a risk quantification model, predicting the judicial risk of the to-be-processed contract based on the risk quantification model, and obtaining a corresponding judicial risk prediction probability, thereby solving the technical problem that the judicial risk cannot be quantitatively predicted due to internal and external data splitting in contract risk control in the prior art. And the compliance efficiency and the anti-risk capability of the contract are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of risk control compliance management, in particular to an intelligent risk control compliance method for a contract. BACKGROUND

[0002] In the current enterprise operation, risk control compliance management is a key link to ensure the stable development of enterprises. With the expansion of business scale, the surge in the number of contracts and the increasingly strict regulatory policies, the traditional risk control compliance management mode gradually relies on manual operation. Manual review of contract terms not only consumes a lot of time, but also is prone to missing risks such as unequal rights and responsibilities and missing default clauses due to human negligence. In the current enterprise contract risk control compliance management, the three core modules of contract management, judicial risk monitoring and compliance knowledge management have been in isolated operation for a long time, forming a significant data fragmentation, so that isolated review is performed on the internal clauses of the contract. This makes it difficult for enterprises to accurately predict the judicial risks that may be faced in the contract performance process, and can only deal with the risks passively after they occur, which seriously affects the safety of contract performance and the efficiency of enterprise risk control compliance, and brings potential economic losses and compliance risks to the enterprise. SUMMARY

[0003] The main purpose of the present application is to provide an intelligent risk control compliance method for a contract, which aims to solve the technical problems of internal and external data fragmentation in contract risk control and inability to quantitatively predict judicial risks in the prior art.

[0004] To achieve the above-mentioned purpose, the present application provides an intelligent risk control compliance method for a contract, which comprises: obtaining a to-be-processed contract, performing structural analysis on the to-be-processed contract, and extracting structured contract data corresponding to the to-be-processed contract; intelligently reviewing the structured contract data based on a preset compliance rule library, and generating risk nodes in the structured contract data; collecting multi-source external data of a preset platform, and extracting contract-related data matching the structured contract data from the multi-source external data, wherein the multi-source external data includes judicial data, administrative data and public opinion data; inputting the structured contract data, the risk nodes and the contract-related data into a risk quantification model, predicting the judicial risk of the to-be-processed contract based on the risk quantification model, and obtaining a corresponding judicial risk prediction probability.

[0005] In an embodiment, the step of performing structural analysis on the to-be-processed contract and extracting the structured contract data corresponding to the to-be-processed contract comprises: when the to-be-processed contract is in the format of an image scan, performing optical character recognition on the to-be-processed contract to obtain corresponding to-be-processed contract text; performing word segmentation on the to-be-processed contract text to obtain a corresponding word segmentation result; performing named entity recognition on the word segmentation result to obtain contract key elements in the to-be-processed contract, the contract key elements including a contract subject, a contract amount, a performance date, and a key obligation clause; performing dependency syntax analysis on the to-be-processed contract text to analyze logical relationships between clauses in the to-be-processed contract text; generating the structured contract data according to the contract key elements and the logical relationships.

[0006] In an embodiment, the intelligent risk control compliance method for the contract further includes: receiving performance documents corresponding to the to-be-processed contract, performing optical character recognition on the performance documents, and extracting actual performance information including an actual performance date and / or an actual payment amount from the performance documents; comparing the actual performance information with preset performance milestones in the structured contract data, the preset performance milestones including payment deadline information and / or payment amount information; generating a warning notification when a deviation between the actual performance information and the performance milestones exceeds a preset tolerance threshold.

[0007] In an embodiment, the step of intelligently reviewing the structured contract data based on a preset compliance rule library to generate risk nodes in the structured contract data includes: loading a compliance rule library through a rule engine, the compliance rule library including three-tuple structured rules of legal provisions, industry standards, and enterprise internal systems; comparing each contract clause in the structured contract data with the legal provisions to obtain a legal provision matching degree corresponding to each contract clause; comparing each contract clause in the structured contract data with the industry standards to obtain an industry standard matching degree corresponding to each contract clause; comparing each contract clause in the structured contract data with the enterprise internal systems to obtain an enterprise system matching degree corresponding to each contract clause; comprehensively integrating the legal provision matching degree, the industry standard matching degree, and the enterprise system matching degree to obtain a compliance matching degree corresponding to each contract clause; determining a contract clause with a compliance matching degree lower than a preset matching degree threshold as the risk node in the structured contract data.

[0008] In an embodiment, the step of collecting multi-source external data of a preset platform includes: when the published content of the preset platform is in a structured format, calling data grabbing of the published content of the preset platform through an application programming interface (API) of the preset platform to obtain the multi-source external data; when the published content of the preset platform is in a semi-structured format, positioning the published content of the preset platform through a cascading style sheets (CSS) selector, extracting the published content, and converting the format of the published content into a structured field to obtain the multi-source external data; when the published content of the preset platform is in an unstructured format, extracting the published content and performing optical character recognition on the published content to obtain the multi-source external data.

[0009] In an embodiment, the step of extracting contract-related data matching the structured contract data from the multi-source external data comprises: determining a contract counterpart of the to-be-processed contract, and determining data associated with the contract counterpart in the judicial data, the administrative data, and the public opinion data as initial contract-related data matching the structured contract data; determining an industry category of the to-be-processed contract, and determining data associated with the industry category in the public opinion data as initial contract-related data matching the structured contract data; inputting the initial contract-related data into a semantic embedding large model for embedding processing to obtain a feature vector of the initial contract-related data; inputting the structured contract data into the semantic embedding large model for embedding processing to obtain a feature vector of the structured contract data; based on the feature vector of the initial contract-related data and the feature vector of the structured contract data, calculating a content matching degree between the initial contract-related data and the structured contract data, and determining the initial contract-related data with a content matching degree greater than a preset threshold as the contract-related data.

[0010] In an embodiment, the step of inputting the structured contract data, the risk node, and the contract-related data into a risk quantification model, and predicting a judicial risk of the to-be-processed contract based on the risk quantification model to obtain a corresponding judicial risk prediction probability comprises: inputting the structured contract data, the risk node, and the contract-related data into a feature embedding layer of the risk quantification model, vectorizing the structured contract data, the risk node, and the contract-related data based on the feature embedding layer to obtain corresponding semantic feature vectors; inputting the semantic feature vector into a bidirectional transformer layer of the risk quantification model, performing self-attention weighted fusion based on the bidirectional transformer layer to obtain a fused feature vector; inputting the fused feature vector into a fully connected layer of the risk quantification model, performing nonlinear transformation on the fused feature vector to obtain the judicial risk prediction probability of the to-be-processed contract.

[0011] In an embodiment, after the step of inputting the structured contract data, the risk node, and the contract-related data into a risk quantification model, predicting the judicial risk of the to-be-processed contract based on the risk quantification model, and obtaining the corresponding judicial risk prediction probability, the method further comprises: obtaining a first early warning threshold and a second early warning threshold, the second early warning threshold being greater than the first early warning threshold; when the judicial risk prediction probability is greater than or equal to the first early warning threshold and less than the second early warning threshold, controlling display of first early warning information, the first early warning information including the judicial risk prediction probability and the to-be-processed contract, and sending the first early warning information to all managers through email; when the judicial risk prediction probability is greater than or equal to the second early warning threshold, controlling display of second early warning information and controlling the user end to issue an alert sound for prompting the second early warning information, and making telephone calls to all the managers through telephone mode.

[0012] In an embodiment, after the step of inputting the structured contract data, the risk node, and the contract-related data into a risk quantification model, predicting the judicial risk of the to-be-processed contract based on the risk quantification model, and obtaining the corresponding judicial risk prediction probability, the method further comprises: in response to an editing instruction, receiving a compliance rule base correction operation input by a user on a visual interface, wherein the compliance rule base correction operation is used to modify the compliance rule base; generating a modified compliance rule base according to the compliance rule base correction operation.

[0013] In an embodiment, after the step of generating the risk node in the structured contract data, the method further comprises: inputting the structured contract data and the risk node into a large language model mounted with the compliance rule base, performing optimized analysis on the structured contract data based on the large language model, and generating a modification suggestion corresponding to the risk node.

[0014] In addition, to achieve the above-mentioned purpose, the present application also provides an intelligent risk control compliance device for a contract, which comprises: A contract structuring module is configured to acquire a to-be-processed contract, perform structural analysis on the to-be-processed contract, and extract structured contract data corresponding to the to-be-processed contract. An intelligent review module is configured to perform intelligent review on the structured contract data based on a preset compliance rule library, and generate a risk node in the structured contract data. A data matching module is configured to collect multi-source external data of a preset platform, and extract contract-related data matching the structured contract data from the multi-source external data, wherein the multi-source external data includes judicial data, administrative data, and public opinion data. A risk prediction module is configured to input the structured contract data, the risk node, and the contract-related data into a risk quantification model, predict judicial risks of the to-be-processed contract based on the risk quantification model, and obtain a corresponding judicial risk prediction probability.

[0015] In addition, to achieve the above object, the present application further provides an intelligent risk control and compliance device for contracts, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the computer program is configured to implement the steps of the intelligent risk control and compliance method for contracts as described above.

[0016] In addition, to achieve the above object, the present application further provides a storage medium, which is a computer readable storage medium, and the storage medium stores a computer program, and the computer program is executed by a processor to implement the steps of the intelligent risk control and compliance method for contracts as described above.

[0017] In addition, to achieve the above object, the present application further provides a computer program product, which comprises a computer program, and the computer program is executed by a processor to implement the steps of the intelligent risk control and compliance method for contracts as described above.

[0018] The one or more technical solutions provided in the application have at least the following technical effects: first, the to-be-processed contract is acquired, the to-be-processed contract is parsed, the structured contract data corresponding to the to-be-processed contract is extracted, the unstructured contract text is converted into standardized data, and a data foundation is laid for subsequent analysis. Based on the preset compliance rule library, the structured contract data is intelligently reviewed, risk nodes in the structured contract data are generated, and the endogenous risks of the contract itself are identified. Multi-source external data of the preset platform is collected, and contract-related data matching the structured contract data is extracted from the multi-source external data, including judicial data, administrative data and public opinion data. The structured contract data, the risk nodes and the contract-related data are input into a risk quantification model, the judicial risk of the to-be-processed contract is predicted based on the risk quantification model, the unified quantification and forward prediction of the risk are realized, the corresponding judicial risk prediction probability is obtained, the deep integration and probabilistic output of the endogenous risk (clause defect) and the exogenous risk (judicial / credit / public opinion) of the contract are realized, the technical problem that the internal and external data are disconnected in the contract risk control in the prior art and the judicial risk cannot be quantitatively predicted is solved, the data islands of the contract, the judicial and the public opinion data are broken, and the compliance efficiency and the anti-risk ability of the contract are improved. The intelligent risk control compliance method of the contract provided in the application extracts the structured contract data by parsing the to-be-processed contract, and converts the unstructured text into a machine-processable standardized field; secondly, the structured contract data is intelligently reviewed based on the preset compliance rule library, and the risk nodes are automatically identified; at the same time, judicial, administrative and public opinion data are collected from multi-source external platforms, and contract-related data matching the structured contract data is accurately extracted, realizing directional association of external risks; and then the structured contract data, the risk nodes and the contract-related data are jointly input into the trained risk quantification model, the judicial risk prediction probability of the to-be-processed contract in litigation and the like is output after comprehensive evaluation, thereby breaking the data barriers between the internal clauses of the contract and the external credit environment, realizing the transition from qualitative judgment to quantitative prediction, and enabling enterprises to actively identify high-risk transactions in the early stage of contract signing or performance, and significantly improving the forward-looking and scientific decision-making of risk control. BRIEF DESCRIPTION OF DRAWINGS

[0019] The accompanying drawings, which are incorporated into and form part of the specification, illustrate embodiments consistent with the present application and, together with the specification, serve to explain the principles of the application.

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the accompanying drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, those skilled in the art can obtain other drawings from these drawings without creative labor.

[0021] Figure 1The schematic diagram of the architecture of the intelligent risk control compliance system of the contract provided in the present application is shown in the figure; Figure 2 The schematic diagram of the process provided in the first embodiment of the intelligent risk control compliance method of the contract in the present application is shown in the figure; Figure 3 The schematic diagram of the module structure of the intelligent risk control compliance device of the contract in the embodiment of the present application is shown in the figure; Figure 4 The schematic diagram of the device structure of the hardware operating environment involved in the intelligent risk control compliance method of the contract in the embodiment of the present application is shown in the figure.

[0022] The object implementation, functional features and advantages of the present application will be further described with reference to the accompanying drawings in conjunction with the embodiments. DETAILED DESCRIPTION

[0023] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application, and are not used to limit the present application.

[0024] In order to better understand the technical solutions of the present application, the following will be described in detail in conjunction with the drawings in the specification and specific embodiments.

[0025] The main solution of the embodiment of the present application is: obtaining a to-be-processed contract, parsing the to-be-processed contract, and extracting structured contract data corresponding to the to-be-processed contract; based on a preset compliance rule library, intelligently reviewing the structured contract data, and generating risk nodes in the structured contract data; collecting multi-source external data of a preset platform, and extracting contract-related data matching the structured contract data from the multi-source external data, the multi-source external data including judicial data, administrative data and public opinion data; inputting the structured contract data, the risk nodes and the contract-related data into a risk quantification model, predicting the judicial risk of the to-be-processed contract based on the risk quantification model, and obtaining a corresponding judicial risk prediction probability.

[0026] In the present embodiment, for the convenience of description, the intelligent risk control compliance system of the contract is taken as the execution subject for elaboration.

[0027] In current business operations, risk control and compliance management are crucial for ensuring stable development. As businesses expand, the number of contracts surges, and regulatory policies become increasingly stringent, traditional risk control and compliance management models increasingly rely on manual operations. Manually reviewing contract terms is not only time-consuming but also prone to oversights, such as misaligned responsibilities or missing breach of contract clauses. Currently, in enterprise contract risk control and compliance management, the three core modules—contract management, judicial risk monitoring, and compliance knowledge management—operate in isolation, resulting in significant data fragmentation and reliance on isolated reviews of internal contract clauses. This makes it difficult for enterprises to accurately predict potential litigation and arbitration risks during contract performance, forcing them to react passively after risks occur. This severely impacts contract performance security and the efficiency of enterprise risk control and compliance, leading to potential economic losses and compliance risks.

[0028] This application provides a solution that involves parsing the contract to be processed, extracting structured contract data, and converting unstructured text into standardized fields that can be processed by machines. Secondly, it intelligently reviews the structured contract data based on a pre-set compliance rule base, automatically identifying risk nodes. Simultaneously, it collects judicial, administrative, and public opinion data from multiple external platforms and accurately extracts contract-related data that matches the structured contract data, achieving targeted correlation with external risks. Finally, it inputs the structured contract data, risk nodes, and contract-related data into a trained risk quantification model to comprehensively assess and output the predicted probability of judicial risks such as litigation arising from the contract to be processed. This breaks down the data barriers between the internal terms of the contract and the external credit environment, achieving a leap from qualitative judgment to quantitative prediction. This allows enterprises to proactively identify high-risk transactions early in the contract signing or performance process, significantly improving the foresight of risk control and the scientific nature of decision-making.

[0029] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or a smart risk control and compliance device for contracts capable of performing the above functions. The following description uses a smart risk control and compliance system for contracts as an example to illustrate this embodiment and the subsequent embodiments.

[0030] The intelligent risk control and compliance method for contracts provided in this application can be applied to intelligent risk control and compliance systems for contracts. For example... Figure 1 As shown, Figure 1 This is a structural diagram of the intelligent risk control and compliance system for contracts. The system's functions are divided according to business logic into: Risk Control Homepage, System Management Module, Contract Management Module, Compliance Management Module, Q&A Section, Legal Litigation, Performance Management, Document Management, System Maintenance, and Mobile Terminal. Table 1 shows a description of each module's function. Table 1. Introduction to the Module Functions of the Intelligent Risk Control and Compliance System for Contracts

[0031] Based on this, the embodiments of the present application provide a smart risk control compliance method for a contract, referring to Figure 2 , Figure 2 The flowchart of the first embodiment of the smart risk control compliance method for a contract of the present application.

[0032] In this embodiment, the smart risk control compliance method for a contract includes steps 201-204: Step 201, obtaining a contract to be processed, performing structured analysis on the contract to be processed, and extracting structured contract data corresponding to the contract to be processed.

[0033] Specifically, the contract to be processed refers to a contract file that needs to be reviewed for risk control or managed for performance by an enterprise, which can be in the form of Word, Portable Document Format (PDF), scanned copy (picture), etc., and has not yet been completed structured processing. The structured contract data refers to the conversion of the contract to be processed into a machine-readable standardized format such as JavaScript Object Notation (JSON), eXtensible Markup Language (XML), or a graph database, which contains key contract elements and their logical relationships, facilitating subsequent program calling and rule matching.

[0034] In some embodiments, after obtaining the to-be-processed contract, the system first determines the format type of the to-be-processed contract. If the to-be-processed contract is an editable electronic document (such as Word or PDF), the text content of the to-be-processed contract can be directly extracted. If the to-be-processed contract is a scanned copy or an image type PDF, an optical character recognition (OCR) engine can be called to perform optical character recognition, and the text structure of the to-be-processed contract can be restored by combining layout analysis technology. Subsequently, the text is processed by natural language processing technology to perform named entity recognition and dependency syntax analysis, and combined with contract domain knowledge templates, key contract elements such as contracting parties, contract amounts, payment conditions, and breach of contract responsibilities are automatically extracted, and finally these information is organized into structured contract data in the form of field-value (such as JSON format). By obtaining the to-be-processed contract, the to-be-processed contract is parsed, and the structured contract data corresponding to the to-be-processed contract is extracted. The original contract (whether an electronic document or a scanned image) that is human-readable but difficult for machines to understand is converted into a standardized data format that can be automatically processed, analyzed, and compared by computers, which can provide a computable, correlatable, and traceable data basis for the entire intelligent risk control process.

[0035] Optionally, the step of structurally analyzing the to-be-processed contract to extract the structured contract data corresponding to the to-be-processed contract comprises: When the to-be-processed contract is in the format of an image scan, performing optical character recognition on the to-be-processed contract to obtain the corresponding to-be-processed contract text; Performing word segmentation processing on the to-be-processed contract text to obtain the corresponding word segmentation result; Performing named entity recognition on the word segmentation result to obtain the contract key elements in the to-be-processed contract, wherein the contract key elements include the contracting parties, the contract amount, the performance date, and the key obligation clauses; Performing dependency syntax analysis on the to-be-processed contract text to analyze the logical relationships between the clauses in the to-be-processed contract text; Generating the structured contract data according to the contract key elements and the logical relationships.

[0036] Specifically, the image scan format refers to a contract file in the form of a picture, such as a PDF scan, a Joint Photographic Experts Group (JPG), a Portable Network Graphics (PNG), a Tagged Image File Format (TIFF), and the like. When the contract to be processed is in the image scan format, the text in the contract to be processed cannot be directly copied, and the text content needs to be recognized through OCR. The contract key elements are the core information with substantive significance in the contract, including the contract subject (such as the name and credit code of Party A and Party B), the contract amount (such as “500 million yuan”), the performance date (such as “delivery before December 31, 2024”), and the key obligation clause (such as “Party B needs to provide a quality certificate”).

[0037] Exemplarily, when the contract to be processed is in the image scan format, a high-precision optical character recognition (OCR) engine is called to recognize the text of the image, and the original document paragraph structure and text order are restored by combining the layout analysis technology to generate an editable contract text to be processed. Subsequently, the Chinese word segmentation processing is performed on the contract text to be processed. The continuous contract text to be processed can be segmented into a word sequence with semantic units by using a word segmentation tool loaded with a professional dictionary in the legal and contract field, such as jieba or LTP. On this basis, a named entity recognition (NER) model based on deep learning can be used to accurately extract contract key elements from the word segmentation result, including the contract subject, the contract amount, the performance date (time node), and the key obligation clause (such as delivery, payment, confidentiality, and the like). At the same time, the dependency syntax analysis is further performed on the contract text to be processed to analyze the syntax and logical relationship between the sentences and the clauses, such as identifying the trigger condition and the legal consequence in the “if … then …” conditional structure, and clarifying the cause-effect relationship between the breach of contract and the performance obligation. The extracted contract key elements and the parsed logical relationship are fused and aligned to organize the structured contract data in a clear field and a standardized structure according to a preset data format (such as JSON Schema). Through the progressive processing of optical character recognition, text processing, entity recognition, logical analysis, and structured generation, the automatic conversion from unstructured image contracts to machine-readable, computable, and inferable structured data is realized, which provides a high-quality and semantically complete input basis for subsequent compliance rule review, risk node identification, and judicial risk quantification, and helps to improve the automation level and accuracy of contract risk control.

[0038] In step 202, the structured contract data is intelligently reviewed based on a preset compliance rule library to generate risk nodes in the structured contract data.

[0039] Specifically, the compliance rule base is a pre-constructed, machine-executable form of compliance knowledge set, covering legal provisions, industry standards, and enterprise internal systems. The risk node is a specific clause or element that violates or deviates from the compliance requirements identified in the structured contract data, such as "no acceptance standard agreed", "payment term exceeds 90 days", "missing intellectual property ownership clause", "only our party is bound in the breach of contract clause, not the other party", etc. Each risk node contains risk type, violation basis, and positioning information.

[0040] In some embodiments, the project type to which the to-be-processed contract belongs can be obtained, and according to the project type to which the to-be-processed contract belongs, such as IT service, engineering construction or procurement, a corresponding rule subset is dynamically loaded from the compliance rule base. Each rule in the compliance rule base is stored in the form of "condition-conclusion". Then, the system inputs the structured contract data into the rule engine, which automatically matches and executes the relevant rules, and performs multi-dimensional comparison between each contract clause in the structured contract data and the rule subset, which can include numerical verification, logical verification and semantic reasoning. Numerical verification can detect whether the penalty proportion exceeds the limit, such as a contract amount of 5 million, a breach of contract of 1.2 million triggering a violation; logical verification can identify contradictory clauses, such as "force majeure exemption" and "waiver of claim rights" coexisting; semantic reasoning can analyze the implied obligations through the BERT model, such as "Party B completes the core module on its own" implying "prohibition of subcontracting". When a contract clause in the structured contract data meets the triggering condition of the rule, the system generates a risk node containing the risk type, violation basis, and contract clause location. For example, if the contract does not specify the ownership of intellectual property rights, it can trigger a "missing key clause" risk node, and if the breach of contract clause only binds our party and not the other party, it is marked as a "unequal rights and responsibilities" risk. By automatically comparing structured contract data with compliance rules, potential legal or commercial risk points are accurately identified, enabling automatic, accurate and interpretable identification of contract risks, upgrading traditional experience-dependent manual review to a standardized, traceable and scalable intelligent review mechanism, and providing a reliable risk input source for contract risk management.

[0041] Optionally, the step of generating risk nodes in the structured contract data based on the preset compliance rule base includes: loading the compliance rule base through the rule engine, the compliance rule base including three tuple structured rules of legal provisions, industry standards, and enterprise internal systems; comparing each contract clause in the structured contract data with the legal provisions to obtain the legal provision matching degree corresponding to each contract clause; The respective contract terms in the structured contract data are compared with industry standards respectively to obtain the industry standard matching degrees corresponding to the respective contract terms. The respective contract terms in the structured contract data are compared with enterprise internal systems respectively to obtain the enterprise system matching degrees corresponding to the respective contract terms. The legal provision matching degree, the industry standard matching degree and the enterprise system matching degree are comprehensively obtained to obtain the compliance matching degrees corresponding to the respective contract terms. The contract terms with a compliance matching degree lower than a preset matching degree threshold are determined as risk nodes in the structured contract data.

[0042] Specifically, the compliance matching degree is a weighted comprehensive score of the matching degrees of the three dimensions of law, industry and enterprise, and can reflect the overall compliance level of the contract terms.

[0043] As an example, a pre-constructed compliance rule library is loaded through a rule engine, which uniformly converts legal provisions, industry standards and enterprise internal systems into machine executable triple structure rules, and marks the source types and applicable scenarios. Subsequently, each contract term (such as payment conditions, breach of contract liability, intellectual property ownership, etc.) in the structured contract data is automatically compared with the three types of rules of legal provisions, industry standards and enterprise internal systems: for legal provisions, keyword matching or semantic similarity calculation can be used to evaluate the legality of the contract terms to generate a legal provision matching degree; for industry standards, the industry association published demonstration text or general practice can be used to obtain the industry standard matching degree through numerical comparison or clause coverage analysis; for enterprise internal systems, strict rule checking can be performed according to company internal control requirements (such as the Contract Management Method) to obtain the enterprise system matching degree. Then, the system can weight and fuse the matching degrees of the three dimensions (legal provision matching degree, industry standard matching degree and enterprise system matching degree) according to the preset weights (for example, legal weight 0.5, industry weight 0.3, enterprise weight 0.2) to calculate the comprehensive compliance matching degree of each contract term. Finally, the contract terms with a compliance matching degree lower than a preset threshold (such as 0.75) are marked as risk nodes, and the contract term risk type, non-compliance basis and original clause position are recorded. Through the core mechanism of multi-dimensional and structured compliance review of the contract, the comprehensiveness and accuracy of risk identification are realized by fusing the three standards of law, industry and enterprise.

[0044] In step 203, multi-source external data of a preset platform is collected, and contract-related data matched with the structured contract data is extracted from the multi-source external data, including judicial data, administrative data and public opinion data.

[0045] Specifically, the preset platform is a pre-configured external data source channel, including a judicial platform, an administrative platform, and an opinion platform. The judicial platform can be China Judgments Online, a list of dishonest persons subject to enforcement, etc. The administrative platform can be a national enterprise credit information public service system, an industry supervision and punishment public service platform, etc. The opinion platform can be a news website, social media, an industry forum, etc. The multi-source external data is external information from different platforms and of different types, including judicial data, administrative data, and opinion data. The judicial data can be judgments, litigation announcements, execution information, etc. The administrative data can be administrative penalty decisions, a list of enterprises with operating abnormalities, qualification and license information, etc. The opinion data can be positive / negative news reports, user reviews, industry comments, etc. The contract-related data is information directly related to the current contract, selected from the multi-source external data, such as the litigation records of the contract party A, the latest supervision and punishment of the industry related to the contract, the latest industry policies, etc.

[0046] In some embodiments, the preset judicial platform, administrative platform, and opinion platform can be connected through an API to automatically collect multi-source external data. The collected multi-source external data can be cleaned, de-duplicated, and standardized (such as unified coding format), and unstructured text can be parsed into structured fields. Further, based on the subject identification (such as credit code), business type, industry category, and other labels in the structured contract data, matching data can be selected through entity association technology to extract the litigation records, administrative penalties, and related opinions of the contract counterpart, and to provide external risk information support for subsequent risk quantification.

[0047] Optionally, the step of collecting multi-source external data from the preset platform includes: When the published content of the preset platform is in a structured format, the published content of the preset platform is grabbed through an API of the preset platform to obtain the multi-source external data; When the published content of the preset platform is in a semi-structured format, the published content of the preset platform is located through a CSS selector, the published content is extracted, and the format of the published content is converted into a structured field to obtain the multi-source external data; When the published content of the preset platform is in an unstructured format, the published content is extracted and optical character recognition is performed on the published content to obtain the multi-source external data.

[0048] Specifically, the structured format is a format in which data is stored according to fixed field specifications, such as JSON, XML, etc. The fields of the structured format have clear meanings and can be directly read by a computer. For example, the "enterprise administrative punishment record" returned by the API contains fields such as "punishment authority", "date", "amount", etc. The semi-structured format is a format in which data has a certain structure but is not strictly specified, such as tables, lists, etc. in HyperText Markup Language (HTML) web pages, which can be extracted by locating tags, such as the row and column data of the "complaint information" table in the web page. The unstructured format is a format without fixed structure, such as scanned documents, images, etc. The content of the unstructured format cannot be directly extracted to obtain field information.

[0049] When collecting multi-source external data of the preset platform, the system can adopt differentiated processing strategies according to the format type of the content published by each platform. For example, when the preset platform provides structured format data, authentication calling can be performed through an open API interface to directly obtain structured data such as judicial decisions, administrative penalties, or enterprise credit information returned in JSON or XML format. When the content of the preset platform is in a semi-structured format, the target elements such as the case parties, the penalty decision date, and the case can be accurately located based on the pre-configured CSS selector, the corresponding text content can be extracted, and the fields can be cleaned, normalized, and structured converted by combining regular expressions and natural language processing techniques. When the content published by the preset platform is in an unstructured format, such as a court announcement published in the form of a picture or a scanned administrative penalty decision, the system downloads the image file, then calls a high-precision optical character recognition engine to convert the text in the image into editable text, and can further extract key risk elements (such as the punished subject, illegal facts, and penalty amount) from the OCR results by template matching or named entity recognition techniques, and finally convert them into structured data. After the collection is completed, the three types of data are de-duplicated and verified, and are summarized to form standardized multi-source external data, thereby ensuring the comprehensive acquisition and format unification of external information such as judicial, administrative, and public opinion, and providing a reliable external data basis for subsequent matching with contract data.

[0050] Optionally, the step of extracting contract-related data matching the structured contract data from the multi-source external data includes: determining the contract counterpart of the to-be-processed contract, and determining the data associated with the contract counterpart in the judicial data, the administrative data, and the public opinion data as initial contract-related data matching the structured contract data; determining the industry category of the to-be-processed contract, and determining the data associated with the industry category in the public opinion data as initial contract-related data matching the structured contract data; inputting the initial contract-related data into a semantic embedding large model for embedding processing to obtain a feature vector of the initial contract-related data; The structured contract data is input into a semantic embedding large model for embedding processing to obtain a feature vector of the structured contract data. Based on the feature vector of the initial contract-related data and the feature vector of the structured contract data, a content matching degree between the initial contract-related data and the structured contract data is calculated, and the initial contract-related data with a content matching degree greater than a preset threshold is determined as the contract-related data.

[0051] Specifically, the semantic embedding large model can be a pre-trained language model such as a bidirectional encoder representation from transformers (BERT), a sentence-level BERT (Sentence-BERT), a chat generative pre-trained transformer (ChatGPT), etc. The text can be mapped to a high-dimensional vector (feature vector) to retain semantic information. The industry category is the industry field to which the contract to be processed belongs, such as "financial technology" and "construction engineering". The industry category is determined based on the contract business content.

[0052] Exemplarily, based on the structured contract data, the contract counterpart of the contract to be processed (such as the full name of the signing enterprise and the unified social credit code) is identified, and all records directly related to the subject are screened from the collected judicial data (such as judgment documents and execution information), administrative data (such as administrative penalties and operation abnormal records) and public opinion data as the first type of initial contract related data according to the contract counterpart. At the same time, the industry category to which the contract to be processed belongs can be determined according to the subject matter or project description of the contract to be processed, and further relevant content such as policy changes, market risks, technical disputes or major events related to the industry can be extracted from the public opinion data as the second type of initial contract related data. Then, the above two types of initial contract related data are spliced into semantic complete text segments respectively, and input into a pre-trained semantic embedding large model such as Sentence-BERT. Each text is mapped to a high-dimensional feature vector by the semantic embedding large model to capture its deep semantic information, and the feature vector of the initial contract related data and the feature vector of the structured contract data are obtained. Then, the cosine similarity between the feature vector of each initial contract related data and the feature vector of the structured contract data is calculated as a quantitative indicator of the content matching degree of the two. A preset threshold (for example, 0.65) is set, and only the initial contract related data with a content matching degree higher than the preset threshold (for example, 0.65) is retained, which is finally determined as the contract related data highly related to the current contract (the contract to be processed). Through the two-stage screening process of “subject + industry primary screening + semantic embedding fine screening”, the external data is first screened by the explicit features, and then screened by the semantic vector, so that the extracted contract related data is highly related to the contract to be processed in the subject, industry and semantic level, and provides accurate external input for the subsequent risk quantification model.

[0053] Step 204, input the structured contract data, risk nodes and contract related data into the risk quantification model, predict the judicial risk of the contract to be processed based on the risk quantification model, and obtain the corresponding judicial risk prediction probability.

[0054] Specifically, the risk quantification model is a machine learning or deep learning model trained by historical data, such as eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), logistic regression, or neural network, which is used to map multi-dimensional risk features to judicial risk probability. Judicial risk refers to the possibility of a contract under processing to occur legal disputes and enter judicial procedures (such as litigation, arbitration, and forced execution) during the performance process, which is the high-order risk form that enterprises are most concerned about. The judicial risk prediction probability is a value between 0 and 1 (such as 0.78) output by the risk quantification model, which represents the probability of the contract to occur judicial events in a specific period (such as within 6 months) in the future, and can be used to support hierarchical early warning and decision-making.

[0055] In some embodiments, structured contract data, risk nodes (such as the absence of liquidated damages, and the imbalance of rights and responsibilities), and contract-related data (such as the counterpart's litigation record and industry negative public opinion) are processed by feature engineering to convert them into a unified dimension numerical feature vector. Then, the feature vector is input into a pre-trained risk quantification model, which has learned the association rules between various features and judicial events based on historical contract performance and judicial dispute data. The risk quantification model outputs a value between 0 and 1, i.e., the judicial risk prediction probability, by forward calculation, which represents the possibility of the contract under processing to occur litigation, arbitration, or forced execution, etc. in a specific period in the future. In addition, after calibrating the output probability, a report containing probability values and key influencing factors can be generated to provide quantitative basis for contract decision-making, and to realize risk visualization and pre-control.

[0056] Based on the intelligent risk control compliance method of the contract provided in the present application, first, the contract to be processed is obtained, the contract to be processed is parsed, the structured contract data corresponding to the contract to be processed is extracted, the unstructured contract text is converted into standardized data, and the data foundation for subsequent analysis is laid. Based on the preset compliance rule library, the structured contract data is intelligently reviewed, the risk nodes in the structured contract data are generated, and the endogenous risks of the contract itself are identified. The multi-source external data of the preset platform is collected, and the contract-related data matching the structured contract data is extracted from the multi-source external data, including judicial data, administrative data and public opinion data. The structured contract data, risk nodes and contract-related data are input into a risk quantification model, the judicial risk of the contract to be processed is predicted based on the risk quantification model, the unified quantification and forward prediction of the risk are realized, the corresponding judicial risk prediction probability is obtained, the deep integration and probabilistic output of the endogenous risk (clause defect) and the exogenous risk (judicial / credit / public opinion) of the contract are realized, and the technical problems of the prior art that the internal and external data are disconnected in the contract risk control and the judicial risk cannot be quantitatively predicted are solved, the data islands of the contract, the judiciary and the public opinion are broken, and the compliance efficiency and the risk resistance of the contract are improved.

[0057] The intelligent risk control compliance method of the contract disclosed in the present application can be applied to an intelligent risk control compliance system of the contract, a deployment mode of the intelligent risk control compliance system of the contract, logical design such as file structure configuration, data access configuration (for example, JDBC) for system deployment, and physical design such as server distribution, License server for system deployment, and system update mode. The hardware environment of the intelligent risk control compliance system of the contract is shown in Table 2: Table 2 Hardware environment of the intelligent risk control compliance system of the contract

[0058] Further, the software environment of the intelligent risk control compliance system of the contract is shown in Table 3: Table 3 Software environment of the intelligent risk control compliance system of the contract

[0059] Further, the technology selection of the intelligent risk control compliance system of the contract is shown in Table 4: Table 4 Technology selection of the intelligent risk control compliance system of the contract

[0060] In some embodiments, the intelligent risk control compliance method of the contract further comprises: receiving the performance document corresponding to the contract to be processed, performing optical character recognition on the performance document, and extracting actual performance information from the performance document, the actual performance information including an actual performance date and / or an actual payment amount; comparing the actual performance information with preset performance milestones in the structured contract data, the preset performance milestones including payment deadline information and / or payment amount information; generating a warning notification when a deviation between the actual performance information and the performance milestones is detected to exceed a preset tolerance threshold.

[0061] Specifically, the performance document is a file generated in the contract performance process to prove the actual performance behavior, such as a bank receipt, an invoice, an acceptance report, a logistics receipt, etc., which can be in a scanned copy or a picture format. The actual performance information is the key execution data extracted from the performance document, such as the actual payment amount, the actual performance date (e.g., “2025-03-15”), reflecting the true performance of the contract. The preset performance milestone is a predetermined node in the structured contract data, such as “50,000 yuan shall be paid before March 1, 2025”, including rigid clauses such as payment deadline information and / or payment amount information. The warning notification is a risk prompt (such as an email, a message or a work order) automatically generated by the system when the deviation exceeds the limit, including deviation details, contract clause basis and suggested measures.

[0062] To further realize the risk management and control of the contract throughout its life cycle, the system also supports dynamic monitoring of the performance stage. As an example, a performance document corresponding to a to-be-processed contract is received, which can be submitted in the form of a scanned copy or an image. Subsequently, a high-precision optical character recognition (OCR) engine is called to perform character recognition on the performance document, and key actual performance information, including actual payment amount and actual performance date, is automatically extracted from the recognition result in combination with a preset field template and a regular expression. Then, the corresponding preset performance milestone is called from the structured contract data generated in the early stage, and the actual performance information is compared with the preset performance milestone item by item to calculate the amount deviation ratio and the time delay days. On this basis, whether there is a substantial deviation is determined according to a preset tolerance threshold (e.g., an amount error of ±2% is allowed, and a time delay of no more than 5 working days is allowed); once a deviation in any dimension exceeds the threshold (e.g., 230,000 yuan is actually paid and 8 days are overdue), the system immediately generates a structured warning notification, the content of which includes the contract number, the deviation details, the contract clause basis and the risk level, and is pushed to relevant business and risk control personnel through a message center, an email or a work order system, realizing the extension from static signing review to dynamic performance monitoring, effectively improving the early discovery and intervention ability of the enterprise on the performance breach risk, thereby ensuring the compliance and safety of contract execution.

[0063] In some embodiments, the step of inputting the structured contract data, the risk node and the contract-related data into a risk quantification model to predict the judicial risk of the to-be-processed contract based on the risk quantification model to obtain a corresponding judicial risk prediction probability includes: inputting the structured contract data, the risk nodes and the contract-related data into a feature embedding layer of a risk quantification model, performing vectorization on the structured contract data, the risk nodes and the contract-related data based on the feature embedding layer, and obtaining corresponding semantic feature vectors; inputting the semantic feature vectors into a bidirectional transformer layer of the risk quantification model, performing self-attention weighted fusion based on the bidirectional transformer layer, and obtaining a fusion feature vector; inputting the fusion feature vector into a fully connected layer of the risk quantification model, performing nonlinear transformation on the fusion feature vector, and obtaining a judicial risk prediction probability of the to-be-processed contract.

[0064] As an example, to achieve accurate quantitative prediction of the judicial risk of a to-be-processed contract, structured contract data, risk nodes and contract-related data are jointly input into a pre-trained risk quantification model. The risk quantification model first performs unified vectorization processing on the three types of heterogeneous inputs (structured contract data, risk nodes and contract-related data) through a feature embedding layer. Specifically, numerical fields can be mapped to dense vectors after normalization, and category or text features are converted through a learnable embedding matrix, and finally semantic feature vectors with consistent dimensions are generated. Subsequently, the semantic feature vectors corresponding to the structured contract data, the semantic feature vectors corresponding to the risk nodes and the semantic feature vectors corresponding to the contract-related data are sent to a bidirectional Transformer layer, which dynamically calculates the correlation weights between features using a self-attention mechanism, realizes context awareness and deep fusion of cross-source features, and outputs high-order representation fusion feature vectors. The fusion feature vector enters the fully connected layer, which can be converted into a judicial risk prediction probability between 0 and 1 through a nonlinear activation function (such as ReLU) and a Sigmoid output layer, representing the possibility of litigation, arbitration or enforcement of the to-be-processed contract within a specific period in the future. Through the deep neural network architecture of the feature embedding layer-bidirectional Transformer layer-fully connected layer of the risk quantification model, the internal defects of the contract and the external credit environment are fused, and the deep neural network captures complex nonlinear risk patterns, thereby improving the prediction accuracy and risk control foresight, and providing a reliable basis for hierarchical early warning and intelligent decision-making.

[0065] In some embodiments, after the step of inputting the structured contract data, the risk nodes and the contract-related data into the risk quantification model, predicting the judicial risk of the to-be-processed contract based on the risk quantification model, and obtaining the corresponding judicial risk prediction probability, the method further comprises: obtaining a first early warning threshold and a second early warning threshold, the second early warning threshold being greater than the first early warning threshold; When the judicial risk prediction probability is greater than or equal to the first warning threshold and less than the second warning threshold, the first warning information is displayed, and the first warning information includes the judicial risk prediction probability and the to-be-processed contract, and the first warning information is sent to all managers by email; When the judicial risk prediction probability is greater than or equal to the second warning threshold, the second warning information is displayed, and the user terminal is controlled to play a warning sound for prompting the second warning information, and all managers are telephoned by telephone.

[0066] Specifically, the first warning threshold and the second warning threshold are preset risk probability critical values, which can be used to divide risk levels and then determine different warning modes. The first warning threshold is low (such as 0.5), and the second warning threshold is high (such as 0.8). When the judicial risk prediction probability is greater than or equal to the first warning threshold and less than the second warning threshold, the risk level of the contract is divided into medium risk. When the judicial risk prediction probability is greater than or equal to the second warning threshold, the risk level of the contract is divided into high risk. The first warning information is prompt information for medium-risk contracts, which contains contract basic information (to-be-processed contract) and risk prediction probability, and is used to remind attention. The second warning information is an urgent warning for high-risk contracts, and the content of the second warning information highlights the risk severity more and triggers stronger intervention measures.

[0067] To achieve hierarchical response and efficient intervention on the judicial risk of the contract, after obtaining the judicial risk prediction probability, an intelligent warning mechanism is further executed. As an example, a preset first warning threshold (such as 0.5) and a second warning threshold (such as 0.8) are obtained, wherein the second warning threshold is higher than the first warning threshold, and is used to divide medium risk and high risk levels. When the judicial risk prediction probability is greater than or equal to the first warning threshold but less than the second warning threshold, it is determined as medium risk, and the first warning information can be automatically popped up on the user terminal. The first warning information content includes the judicial risk prediction probability and the to-be-processed contract, and the email service is called synchronously to push the first warning information to the enterprise mailbox of all managers. When the judicial risk prediction probability is greater than or equal to the second warning threshold, it is determined as high risk. In addition to displaying the second warning information in a conspicuous manner on the user terminal, the client can also play a warning sound (such as a buzzer prompt sound) to attract the attention of the operator, and at the same time, the integrated voice communication platform (such as an enterprise call center or a cloud phone) can automatically dial the reserved phone numbers of all managers to perform voice broadcast notification, ensuring that critical risks are timely perceived and processed. According to the level of the judicial risk prediction probability, a differentiated warning strategy is implemented to ensure that risk information reaches relevant personnel in a timely and effective manner, avoiding low-risk interference and ensuring that high-risk is not missed, thereby improving the response speed and collaborative processing capability of enterprises to major contract risks.

[0068] In some embodiments, after the step of inputting the structured contract data, risk nodes and contract-related data into the risk quantification model, predicting the judicial risk of the to-be-processed contract based on the risk quantification model, and obtaining the corresponding judicial risk prediction probability, the method further comprises: In response to the editing instruction, receiving a compliance rule base correction operation input by the user on the visualization interface, wherein the compliance rule base correction operation is used to modify the compliance rule base; According to the compliance rule base correction operation, a modified compliance rule base is generated.

[0069] Specifically, the editing instruction is a modification operation trigger signal initiated by the user on the system interface, such as clicking the "Edit Rule" button or dragging to adjust the rule weight, indicating the user's intention to update the compliance rule base. The visualization interface is a graphical operation panel provided by the system, which can visually display the compliance rule content in the form of tables, forms, flowcharts, etc. The compliance rule base correction operation is a specific modification behavior performed by the user, including but not limited to: adding / deleting rules, adjusting rule conditions, modifying rule sources (laws / industries / companies), setting applicable contract types, or adjusting rule priorities, etc.

[0070] As an example, when the user initiates an editing instruction in the visualization management interface (such as a Web background), such as clicking the "Rule Edit" button, the system displays the rule base content on the visualization interface after responding. The user performs a correction operation on the visualization interface. The compliance rule base correction operation can include adding, deleting, or modifying a structured rule (for example, adjusting "penalty ratio not less than 5%" to "not less than 8%"), changing the contract type applicable to the rule, adding a rule (such as entering "service contract needs to specify acceptance criteria"), or disabling rules with high false positive rates, etc. In addition, the correction operation can also be subjected to syntax and logic checking (such as numerical reasonableness, rule conflict detection), and after passing the checking, the changed content is written into the compliance rule base database persistently, and a modified compliance rule base with a version number is generated. At the same time, the rule engine is automatically notified to reload the latest rules to ensure that subsequent contract intelligent review is based on the updated standards.

[0071] In addition, to strengthen the collaborative governance and continuous evolution of compliance knowledge, a compliance interactive platform based on Elasticsearch is constructed, and is deeply linked with the intelligent contract review full process. The compliance interactive platform constructs a structured compliance knowledge graph from multi-source data such as laws and regulations, industry standards, enterprise systems and historical contract cases, supports millisecond-level efficient retrieval according to keywords, clause types or risk labels, and when controversial clauses (such as differences in the scope of “force majeure”) appear in the contract review process, legal, business or risk control personnel can initiate online discussions on the platform, collaboratively edit, mark and upload supporting materials around specific clauses. The compliance interactive platform supports real-time collaboration of multiple people and provides a voting function to facilitate quick consensus. If the conclusion of the discussion has universality, it can be converted into a structured compliance rule or a typical case after administrator review, and automatically injected into the compliance rule library or knowledge graph; the new rule is immediately loaded by the rule engine for subsequent contract review, realizing a closed-loop management from practice disputes to collective decision-making to knowledge sedimentation and intelligent application, and improving the agility and professionalism of compliance response.

[0072] In some embodiments, after the step of generating the risk node in the structured contract data, further comprising: inputting the structured contract data and the risk node into a large language model mounted with the compliance rule library, performing optimized analysis on the structured contract data based on the large language model, and generating a modification suggestion corresponding to the risk node.

[0073] Specifically, after generating the risk node in the structured contract data, the structured contract data and the risk node are further input into a large language model mounted with the compliance rule library. The large language model adopts a Retrieval-Augmented Generation (RAG) architecture, and according to the content of the risk node, the most relevant legal provisions, industry standards and enterprise internal systems can be retrieved from the vectorized stored compliance rule library in real time; then, the retrieved compliance basis (legal provisions, industry standards and enterprise internal systems) and the contract context are jointly structured to guide the large language model to generate a modification suggestion that is semantically coherent, compliant and directly replaceable. Through optimized analysis, the large language model generates a specific modification suggestion for each risk node. The generated modification suggestion can be displayed side by side with the original clause in the user interface and supports one-key adoption. By converting the risk node into a feasible revision plan, the compliance of the suggestion is ensured, the threshold of manual revision is reduced, and the efficiency of contract compliance is improved.

[0074] It should be noted that the above examples are only for understanding the present application and do not constitute a limitation on the intelligent risk control compliance method of the present application. Further simple transformations based on this technical concept are within the protection scope of the present application.

[0075] The present application also provides an intelligent risk control compliance device for a contract, which is described with reference toFigure 3 The intelligent risk control compliance device for a contract comprises: A contract structuring module 301 is configured to acquire a to-be-processed contract, perform structural analysis on the to-be-processed contract, and extract structured contract data corresponding to the to-be-processed contract; An intelligent review module 302 is configured to perform intelligent review on the structured contract data based on a preset compliance rule library, and generate a risk node in the structured contract data; A data matching module 303 is configured to collect multi-source external data of a preset platform, and extract contract-related data matching the structured contract data from the multi-source external data, wherein the multi-source external data comprises judicial data, administrative data, and public opinion data; A risk prediction module 304 is configured to input the structured contract data, the risk node, and the contract-related data into a risk quantification model, predict a judicial risk of the to-be-processed contract based on the risk quantification model, and obtain a corresponding judicial risk prediction probability.

[0076] The intelligent risk control compliance device for a contract provided in the present application adopts the intelligent risk control compliance method for a contract in the above embodiment, and can solve the technical problem of the inability to quantitatively predict a judicial risk due to the internal and external data being split in the contract risk control in the prior art. Compared with the prior art, the intelligent risk control compliance device for a contract provided in the present application has the same beneficial effects as the intelligent risk control compliance method for a contract provided in the above embodiment, and other technical features in the intelligent risk control compliance device for a contract are the same as the features disclosed in the above embodiment method, which will not be described herein.

[0077] The present application provides an intelligent risk control compliance device for a contract, which comprises at least one processor and a memory in communication connection with the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the intelligent risk control compliance method for a contract in the above embodiment one.

[0078] Reference will be made to the following description of the drawings to understand the intelligent risk control compliance device for a contract. Figure 4 which shows a structural schematic diagram of an intelligent risk control compliance device for a contract suitable for use to implement the embodiments of the present application. The intelligent risk control compliance device for a contract in the embodiments of the present application can include, but is not limited to, mobile terminals such as notebook computers, Portable Application Descriptions (PADs), vehicle-mounted terminals (for example, vehicle-mounted navigation terminals), and the like, and fixed terminals such as digital TVs, desktop computers, and the like. Figure 4 The illustrated intelligent risk control compliance device for a contract is merely an example, and should not impose any limitation on the functions and use range of the embodiments of the present application.

[0079] As Figure 4As shown, the smart contract risk control and compliance device can include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.) that can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) 1002 or loaded from a storage device 1003 into a random access memory (RAM) 1004. Various programs and data required for the operation of the smart contract risk control and compliance device are also stored in the random access memory 1004. The processing device 1001, the read-only memory 1002, and the random access memory 1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the input / output interface 1006: input devices 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; the storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the smart contract risk control and compliance device to communicate with other devices wirelessly or by wire to exchange data. Although the smart contract risk control and compliance device with various systems is shown in the figure, it should be understood that all the shown systems are not required to be implemented or possessed. More or fewer systems can be alternatively implemented or possessed.

[0080] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product comprising a computer program carried on a computer readable medium, the computer program containing program codes for executing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network through the communication device, or installed from the storage device 1003, or installed from the read-only memory 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the methods of the embodiments disclosed in the present application are performed.

[0081] The smart contract risk control and compliance device provided by the present application adopts the smart contract risk control and compliance method in the above-mentioned embodiments, and can solve the technical problems of internal and external data fragmentation and inability to quantitatively predict judicial risks in contract risk control in the prior art. Compared with the prior art, the smart contract risk control and compliance device provided by the present application has the same beneficial effects as the smart contract risk control and compliance method provided by the above-mentioned embodiments, and other technical features in the smart contract risk control and compliance device are the same as the features disclosed in the previous embodiment method, which will not be repeated here.

[0082] It should be understood that various parts of the present disclosure can be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0083] The above description is merely illustrative of the application and not restrictive.

[0084] The present application provides a computer readable storage medium having stored thereon computer readable program instructions (i.e., computer programs) for performing the smart risk control compliance method of a contract in the above-described embodiments.

[0085] The computer readable storage medium provided by the present application may, for example, be a U disk, but is not limited to an electric, magnetic, optical, electromagnetic, infrared, or semiconductor system, system, or device, or any combination thereof. More specific examples of the computer readable storage medium can include, but are not limited to, an electric connection with one or more conductive wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present embodiment, 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, system, or device. The program code contained on the computer readable storage medium can be transmitted by any suitable medium, including but not limited to an electric wire, an optical cable, an RF (Radio Frequency), etc., or any suitable combination thereof.

[0086] The above-described computer readable storage medium can be included in the smart risk control compliance device of a contract; or can exist separately and not be assembled into the smart risk control compliance device of a contract.

[0087] The computer readable storage medium described above carries one or more programs, when the one or more programs are executed by the intelligent risk control compliance device of the contract, the intelligent risk control compliance device of the contract: obtaining a to-be-processed contract, parsing the to-be-processed contract, and extracting structured contract data corresponding to the to-be-processed contract; based on a preset compliance rule library, intelligently reviewing the structured contract data, and generating a risk node in the structured contract data; collecting multi-source external data of a preset platform, and extracting contract-related data matching the structured contract data from the multi-source external data, the multi-source external data including judicial data, administrative data, and public opinion data; inputting the structured contract data, the risk node, and the contract-related data into a risk quantification model, predicting the judicial risk of the to-be-processed contract based on the risk quantification model, and obtaining a corresponding judicial risk prediction probability.

[0088] Computer program code for carrying out operations of the present application can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0089] The flow diagrams and the block diagrams in the drawings are illustrations of architectures, functionalities, and operations of possible implementations of systems, methods, and computer program products according to various embodiments of present application. In this regard, each block in the flow diagrams or block diagrams can represent a module, a segment, or a portion of code, which comprises one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and / or flow diagrams, and combinations thereof, can be implemented by special purpose hardware-based systems that perform the specified functions or operations, or combinations of special purpose hardware and computer instructions.

[0090] The modules described in the embodiments of the present application can be implemented in the form of software or in the form of hardware. In some cases, the name of the module does not constitute a limitation on the module itself.

[0091] The readable storage medium provided by the present application is a computer readable storage medium, which stores computer readable program instructions (i.e., a computer program) for executing the smart risk control compliance method of the contract described above, and can solve the technical problems of internal and external data fragmentation and inability to quantitatively predict judicial risks in contract risk control in the prior art. Compared with the prior art, the computer readable storage medium provided by the present application has the same beneficial effects as the smart risk control compliance method of the contract provided by the above embodiments, and will not be described here.

[0092] The present application also provides a computer program product comprising a computer program, which, when executed by a processor, implements the steps of the smart risk control compliance method of the contract as described above.

[0093] The computer program product provided by the present application can solve the technical problems of internal and external data fragmentation and inability to quantitatively predict judicial risks in contract risk control in the prior art. Compared with the prior art, the computer program product provided by the present application has the same beneficial effects as the smart risk control compliance method of the contract provided by the above embodiments, and will not be described here.

[0094] The above only describes some embodiments of the present application, and does not limit the patent scope of the present application. Any equivalent structural transformation, direct / indirect application in other related technical fields based on the technical concept of the present application, and the contents of the present application specification and drawings are included in the patent protection scope of the present application.

Claims

1. A smart risk control and compliance method for contracts, characterized in that, The intelligent risk control and compliance methods for the contract include: Obtain the contract to be processed, perform structured parsing on the contract to be processed, and extract the structured contract data corresponding to the contract to be processed; The structured contract data is intelligently reviewed based on a preset compliance rule base to generate risk nodes in the structured contract data. Collect multi-source external data from a preset platform, and extract contract-related data that matches the structured contract data from the multi-source external data. The multi-source external data includes judicial data, administrative data, and public opinion data. The structured contract data, the risk nodes, and the contract-related data are input into the risk quantification model. Based on the risk quantification model, the judicial risk of the contract to be processed is predicted, and the corresponding judicial risk prediction probability is obtained.

2. The intelligent risk control and compliance method for contracts as described in claim 1, characterized in that, The step of performing structured parsing on the contract to be processed and extracting the structured contract data corresponding to the contract to be processed includes: When the contract to be processed is in the format of an image scan, optical character recognition is performed on the contract to be processed to obtain the corresponding contract text. The contract text to be processed is segmented into words to obtain the corresponding segmentation results; Named entity recognition is performed on the word segmentation results to obtain the key contract elements in the contract to be processed. The key contract elements include the contract parties, contract amount, performance date, and key obligation clauses. Dependency parsing is performed on the contract text to be processed to parse the logical relationships between the clauses in the contract text. The structured contract data is generated based on the key elements of the contract and the logical relationships.

3. The intelligent risk control and compliance method for contracts as described in claim 1, characterized in that, The intelligent risk control and compliance methods for the contract also include: Receive the performance documents corresponding to the contract to be processed, perform optical character recognition on the performance documents, and extract the actual performance information from the performance documents, including the actual performance date and / or the actual payment amount; The actual performance information is compared with the preset performance milestones in the structured contract data, wherein the preset performance milestones include payment deadline information and / or payment amount information; When the deviation between the actual performance information and the performance milestone is detected to exceed a preset tolerance threshold, an early warning notification is generated.

4. The intelligent risk control and compliance method for contracts as described in claim 1, characterized in that, The step of intelligently reviewing the structured contract data based on a preset compliance rule base and generating risk nodes in the structured contract data includes: The compliance rule library is loaded through the rule engine. The compliance rule library contains a triplet structured rule library containing legal provisions, industry standards, and internal corporate regulations. Each contract clause in the structured contract data is compared with the legal provisions to obtain the legal provision matching degree corresponding to each contract clause; Each contract clause in the structured contract data is compared with the industry standard to obtain the industry standard matching degree corresponding to each contract clause; Each contract clause in the structured contract data is compared with the company's internal regulations to obtain the matching degree of the company's regulations for each contract clause; By combining the matching degree of the legal provisions, the matching degree of the industry standards, and the matching degree of the corporate system, the compliance matching degree of each contract clause is obtained; Contract clauses with a compliance matching degree lower than a preset matching degree threshold are identified as risk nodes in the structured contract data.

5. The intelligent risk control and compliance method for contracts as described in claim 1, characterized in that, The steps for collecting multi-source external data from the preset platform include: When the content published on the preset platform is in a structured format, the data published on the preset platform is retrieved by calling the API of the preset platform to obtain the multi-source external data; When the published content on the preset platform is in a semi-structured format, the published content on the preset platform is located using a CSS selector, the published content is extracted, and the format of the published content is converted into structured fields to obtain the multi-source external data; When the published content on the preset platform is in an unstructured format, the published content is extracted and optical character recognition is performed on the published content to obtain the multi-source external data.

6. The intelligent risk control and compliance method for contracts as described in claim 1, characterized in that, The step of extracting contract-related data matching the structured contract data from the multi-source external data includes: The counterparty to the contract to be processed is identified, and the data associated with the counterparty in the judicial data, administrative data, and public opinion data are identified as the initial contract-related data to be matched with the structured contract data; The industry category of the contract to be processed is determined, and the data in the public opinion data associated with the industry category is identified as the initial contract-related data that matches the structured contract data; The initial contract-related data is input into a semantic embedding model for embedding processing to obtain the feature vector of the initial contract-related data. The structured contract data is input into the semantic embedding model for embedding processing to obtain the feature vector of the structured contract data; Based on the feature vectors of the initial contract-related data and the structured contract data, the content matching degree between the initial contract-related data and the structured contract data is calculated, and the initial contract-related data with a content matching degree greater than a preset threshold is determined as the contract-related data.

7. The intelligent risk control and compliance method for contracts as described in claim 1, characterized in that, The step of inputting the structured contract data, the risk nodes, and the contract-related data into a risk quantification model, and predicting the judicial risk of the contract to be processed based on the risk quantification model to obtain the corresponding judicial risk prediction probability includes: The structured contract data, the risk nodes, and the contract-related data are input into the feature embedding layer of the risk quantification model. Based on the feature embedding layer, the structured contract data, the risk nodes, and the contract-related data are vectorized to obtain the corresponding semantic feature vectors. The semantic feature vector is input into the bidirectional transformer layer of the risk quantification model, and self-attention weighted fusion is performed based on the bidirectional transformer layer to obtain the fused feature vector. The fused feature vector is input into the fully connected layer of the risk quantification model, and a nonlinear transformation is performed on the fused feature vector to obtain the judicial risk prediction probability of the contract to be processed.

8. The intelligent risk control and compliance method for contracts as described in claim 1, characterized in that, After the step of inputting the structured contract data, the risk nodes, and the contract-related data into the risk quantification model, predicting the judicial risk of the contract to be processed based on the risk quantification model, and obtaining the corresponding judicial risk prediction probability, the method further includes: Obtain a first warning threshold and a second warning threshold, wherein the second warning threshold is greater than the first warning threshold; When the predicted probability of judicial risk is greater than or equal to the first warning threshold and less than the second warning threshold, the system displays the first warning information, which includes the predicted probability of judicial risk and the pending contract, and sends the first warning information to all administrators via email. When the predicted probability of judicial risk is greater than or equal to the second warning threshold, the system controls the display of the second warning information and controls the user terminal to emit a warning sound to prompt the second warning information, and also controls the system to notify all the management personnel by telephone.

9. The intelligent risk control and compliance method for contracts as described in claim 1, characterized in that, After the step of inputting the structured contract data, the risk nodes, and the contract-related data into the risk quantification model, predicting the judicial risk of the contract to be processed based on the risk quantification model, and obtaining the corresponding judicial risk prediction probability, the method further includes: In response to an editing command, the system receives a compliance rule base modification operation input by the user on the visual interface, wherein the compliance rule base modification operation is used to modify the compliance rule base; Based on the compliance rule base correction operation, the corrected compliance rule base is generated.

10. The intelligent risk control and compliance method for contracts as described in claim 1, characterized in that, Following the step of generating risk nodes in the structured contract data, the method further includes: The structured contract data and the risk nodes are input into a large language model that is mounted on the compliance rule base. Based on the large language model, the structured contract data is optimized and analyzed to generate modification suggestions corresponding to the risk nodes.

Images