A contract automation review system and method based on multi-agent collaboration

Abstract

The present invention belongs to the field of artificial intelligence technology and provides a contract automation review system and method based on multi-agent collaboration, including configuring contract review description information and generating a contract review request; receiving the contract review request, responding to the contract review request and performing semantic analysis on the contract review description information to extract key review elements; calling a pre-trained intention recognition agent in an agent library, using the intention recognition agent to sequentially perform keyword, semantic, and logical multi-level matching on the key review elements with a pre-built review rule library to obtain the review intent; based on the attributes of the review intent, matching and searching the corresponding pre-trained agent from the agent library and calling the called agent to collaboratively review the contract content to obtain the review result. The system can perform automated and accurate review of contracts based on multi-agent collaboration.

Description

Technical Field

[0001] The present invention belongs to the field of artificial intelligence technology, and in particular relates to a contract automation review system and method based on multi-agent collaboration. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] In the current contract review field, although big models can perform keyword retrieval and clause comparison, assisting manual review and improving contract review efficiency, the following issues still exist: Currently, big model review often relies on keyword matching or preset rule libraries. Contracts have many rules, but the review rules are simple and lack a deep understanding of the contract's semantics. When faced with complex legal clauses and industry terminology, they are prone to missed detections and misjudgments. This results in poor accuracy of intelligent review using big model agents, making it impossible to conduct precise reviews. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention provides a contract automation review system and method based on multi-agent collaboration, which can automatically and accurately review contracts based on multi-agent collaboration.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] The first aspect of the present invention provides a contract automation review system based on multi-agent collaboration.

[0007] In one or more embodiments, a multi-agent collaborative contract automation review system is provided, including:

[0008] The front-end configuration module is used to: configure contract review description information and generate contract review requests;

[0009] Backend processing module, which is used to:

[0010] Receive a contract review request, respond to the contract review request, and perform semantic analysis on the contract review description information to extract key review elements;

[0011] Call the pre-trained intent recognition agent in the agent library, and use the intent recognition agent to match the key audit elements with the pre-built audit rule library in terms of keywords, semantics and logic at multiple levels to obtain the audit intent;

[0012] Based on the attributes of the audit intent, the corresponding pre-trained agent is matched and called from the agent library, and the called agent is used to collaboratively audit the contract content to obtain the audit results.

[0013] As an implementation method, in the back-end processing module, the process of obtaining the audit intent is as follows:

[0014] Match key audit elements with keywords in the pre-built audit rule library to preliminarily locate the scope of rule entries in the audit rule library;

[0015] Perform semantic identification on key audit elements, analyze the contextual relationship of key audit elements, determine whether they are consistent with the semantics of rules within the scope of the initially located rule entries, and obtain rule entries with consistent semantics;

[0016] Conduct a logical relationship analysis on key audit elements, and then, based on the results of the logical relationship analysis, screen the final audit rules from the rule entries that match the rule semantics, and use the final audit rules as the audit intent.

[0017] As an implementation manner, in the back-end processing module, after obtaining the audit intention, the following is further included:

[0018] Evaluate the audit intent to determine whether it is consistent with the contract audit description information, and revise the audit intent that is inconsistent with the contract audit description information.

[0019] As an implementation method, the back-end processing module further includes evaluating the audit results to determine whether the audit results meet the set requirements.

[0020] As an implementation mode, in the back-end processing module, when the audit result meets the set requirements, the audit is ended; when the audit result does not meet the set requirements, the audit intention is corrected and supplemented until the audit result meets the set requirements.

[0021] As an implementation manner, in the back-end processing module, the attributes of the audit intention include auditing the legal attributes in the contract and auditing the financial attributes in the contract.

[0022] As an implementation method, in the back-end processing module, when the attribute of the audit intention is to audit the legal attribute in the contract, the pre-trained legal agent is matched, searched and called from the agent library; when the attribute of the audit intention is to audit the financial attribute in the contract, the pre-trained financial agent is matched, searched and called from the agent library.

[0023] As an embodiment, the multi-agent collaborative contract automation review system further includes:

[0024] The audit result feedback module is used to feed back the audit results to the front-end configuration module.

[0025] The second aspect of the present invention provides a contract automation review method based on multi-agent collaboration.

[0026] In one or more embodiments, a multi-agent collaborative contract automation review method, executed in a backend processing module, includes the following steps:

[0027] Receive a contract review request generated by configuring the contract review description information by the front-end configuration module, respond to the contract review request and perform semantic analysis on the contract review description information to extract key review elements;

[0028] Call the pre-trained intent recognition agent in the agent library, and use the intent recognition agent to match the key audit elements with the pre-built audit rule library in terms of keywords, semantics and logic at multiple levels to obtain the audit intent;

[0029] Based on the attributes of the audit intent, the corresponding pre-trained agent is matched and called from the agent library, and the called agent is used to collaboratively audit the contract content to obtain the audit results.

[0030] A third aspect of the present invention provides an electronic device.

[0031] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps of the above-described method for automated contract review based on multi-agent collaboration are implemented.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] (1) The present invention generates a contract review request based on the configured contract review description information and processes the contract review request. First, the contract review description information is semantically parsed to extract key review elements. Then, the intent recognition agent is called from the agent library to perform multi-level matching of key review elements and the review rule library. Then, based on the attributes of the review intent, the corresponding pre-trained agent is matched and called from the agent library. The called agent is used to collaboratively review the contract content. The collaborative scheduling of multiple agents in the agent library is used to achieve automated and accurate review of the contract content.

[0034] (2) The present invention enables contract review to cover multiple fields such as law, finance, and business through multi-agent collaboration and precise intent recognition; the use of pre-trained agents for collaborative review reduces human omissions and subjective judgment errors, and can accurately identify risk clauses in contracts, helping enterprises avoid potential economic and legal risks; and through the combination of flexible configuration of the front-end configuration module and dynamic response of the back-end processing module, the system can easily cope with contract reviews of different industries and different levels of complexity. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0036] Figure 1 Schematic diagram of the structure of a multi-agent collaborative contract automation review system according to an embodiment of the present invention;

[0037] Figure 2 This is a flowchart of a multi-agent collaborative contract automation review method according to an embodiment of the present invention;

[0038] Figure 3 is a schematic diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0039] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0040] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0041] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0042] Figure 1 This is a schematic diagram of the structure of a contract automation review system based on multi-agent collaboration in an embodiment of the present invention. Figure 1 The automated contract review system based on multi-agent collaboration in this embodiment may include the following modules:

[0043] The front-end configuration module 101 is used to configure contract review description information and generate a contract review request.

[0044] In this embodiment, the front-end configuration module 101 can be a user-end device, such as an electronic device with input functionality. The user can configure contract review description information through the front-end configuration module 101. The contract review description information here includes, but is not limited to, price clauses and delivery date clauses in the contract, as well as review criteria, such as whether the price is within a reasonable range and whether the delivery date complies with industry practices.

[0045] During implementation, the front-end configuration module 101 also provides a visual configuration tool that supports users in entering contract review requirements in a variety of ways and converts them into structured review descriptions. This flexible configuration method breaks the limitations of the traditional review model, where requirements are vague and single-minded. It can accurately match users' diverse review requirements and ensure the accuracy of review direction from the source.

[0046] The backend processing module 102 is configured to perform the following steps:

[0047] Step a: Receive a contract review request, respond to the contract review request, perform semantic analysis on the contract review description information, and extract key review elements.

[0048] In step a, the process of semantically parsing the contract review description information is an existing technology, and an existing semantic model, such as a BERT model or a GPT model, can be used.

[0049] The extracted key audit elements include but are not limited to the audit objects and audit standards.

[0050] Step b: Call the pre-trained intention recognition agent in the agent library, and use the intention recognition agent to perform multi-level matching of keywords, semantics and logic on the key audit elements with the pre-built audit rule library to obtain the audit intent.

[0051] It should be noted here that the agent library pre-stores an intent recognition agent, which is pre-trained using an existing large model.

[0052] In this embodiment, the audit rule library is constructed through multi-channel data collection and professional knowledge integration. On the one hand, a large number of historical contract audit cases are collected and analyzed to analyze common audit points and standards. For example, in construction project contracts, the audit rules for clauses such as payment methods and compensation for construction delays are also analyzed. On the other hand, legal, financial, and industry experts are invited to participate in the formulation of rules, incorporating legal and regulatory requirements, industry norms, and internal corporate systems.

[0053] In this embodiment, the audit rule base has a dynamic update mechanism, which can timely update and improve the audit rules according to changes in laws and regulations, adjustments to industry standards, and changes in the business needs of enterprises to ensure the timeliness and accuracy of the rule base.

[0054] In step b, the process of obtaining the audit intent is:

[0055] Step b1: Match the key audit elements with the keywords in the pre-built audit rule library to preliminarily locate the scope of the rule entries in the audit rule library.

[0056] It should be noted here that the expression of key audit elements is consistent with the keyword expression in the audit rule library, which can improve the accuracy of matching.

[0057] Step b2: Perform semantic identification on the key audit elements, analyze the contextual relationship of the key audit elements, determine whether they are consistent with the rule semantics within the scope of the initially located rule entries, and obtain the rule entries with consistent rule semantics.

[0058] In the process of semantic matching, existing large models such as the BERT model are used for semantic understanding and recognition, to understand the semantic meaning of the audit elements, analyze their contextual relationships, and determine whether they are consistent with the semantics of the rules in the rule base.

[0059] Step b3: Perform a logical relationship analysis on the key audit elements, and then, based on the results of the logical relationship analysis, select the final audit rules from the rule entries that match the rule semantics, and use the final audit rules as the audit intent.

[0060] In step b3, logical relationships include but are not limited to conditional relationships, causal relationships, etc. For example, a decision tree model can be used to analyze the logical relationships of key audit elements, or existing methods such as the logical framework method can be used to analyze the logical relationships.

[0061] This embodiment accurately determines audit intent through multi-level matching based on keywords, semantics, and logic. This process overcomes the limitation of a single audit model that cannot adapt to diverse contract types and complex requirements, enabling the system to "understand" audit requirements. Based on this, matching agents are retrieved from the agent library based on the intent recognition results. Each agent specializes in a specific audit area, avoiding the professional limitations and subjective bias of manual audits, achieving precise allocation of audit resources, and comprehensively ensuring the accuracy and efficiency of audit work.

[0062] In one or more embodiments, after obtaining the audit intention, the backend processing module further includes:

[0063] Evaluate the audit intent to determine whether it is consistent with the contract audit description information, and revise the audit intent that is inconsistent with the contract audit description information.

[0064] For example, if it is manually discovered that the audit intention is inconsistent with the overall content of the contract, business background, etc., or deviates from the audit requirements input by the user, the audit intention can be manually judged and scored, thus forming a feedback mechanism. Users can correct their intentions by tracing back the audit element parsing process, re-matching rules, etc. by scoring the audit intention.

[0065] At the same time, if new audit requirements or potential intentions that are not clearly expressed are discovered during the audit process, the intentions can be supplemented based on the audit rule base and relevant knowledge to ensure a comprehensive and accurate understanding of the contract audit requirements.

[0066] Step c: Based on the attributes of the audit intent, match and search and call the corresponding pre-trained agent from the agent library, and use the called agent to collaboratively audit the contract content to obtain the audit results.

[0067] In one or more embodiments, the back-end processing module further includes evaluating the audit results to determine whether the audit results meet the set requirements.

[0068] In the back-end processing module, when the audit result meets the set requirements, the audit is terminated; when the audit result does not meet the set requirements, the audit intention is revised and supplemented until the audit result meets the set requirements. This can ensure the accuracy of the audit results.

[0069] In the backend processing module, the audit intent attributes include the legal attributes of the contract and the financial attributes of the contract. When the audit intent attributes are the legal attributes of the contract, a pre-trained legal agent is searched and invoked from the agent library; when the audit intent attributes are the financial attributes of the contract, a pre-trained financial agent is searched and invoked from the agent library.

[0070] Each audit intent is assigned a fixed and unique agent ID in the agent library, allowing you to select a matching agent from multiple agent libraries to perform contract audits. Each agent is trained to possess audit capabilities specific to a specific domain. For example, a legal agent excels at reviewing legal risk clauses in contracts, while a financial agent accurately reviews financial data and clauses within contracts. The selected agent is then called upon to review the contract content and obtain the audit results.

[0071] Taking a business cooperation contract as an example, we first perform semantic analysis on the contract review request submitted by the user to identify keywords and key phrases, such as "legal compliance" and "financial terms". Combined with the preset review rule library, we determine that the review intent includes legal compliance and financial terms.

[0072] The backend processing module invokes the legal and financial agents in the agent library. The legal agent is built based on a 72B large-scale model covering the legal field. Its training data samples include over one million articles of legal regulations, judicial interpretations, typical cases, and legal academic literature. The legal agent receives the full text of the contract and the review requirements conveyed by the intent recognition agent as input. After in-depth analysis and reasoning by the model, it outputs structured data, including the content of clauses posing legal risks, the risk type, the specific legal provisions violated, and the risk level assessment.

[0073] The Financial Agent in the Agent Library is built based on a 72-byte large model. Its training data includes 800,000 pieces of corporate financial statements, tax regulations, financial analysis reports, and industry financial indicator data. The Financial Agent's input is the financial clauses in the contract and risk information provided by the Legal Agent. Its output is quantified financial impact data, including the range of possible fines, estimated compensation liability, and analysis of the impact on financial indicators.

[0074] according to Figure 1 , the contract automation review system based on multi-agent collaboration also includes:

[0075] The audit result feedback module 103 is used to feed back the audit result to the front-end configuration module 101 .

[0076] For example, the audit result feedback module 103 feeds back the audit results in a structured form to the front-end configuration module 101. This interactive collaboration process is achieved through message passing and data sharing. Each intelligent agent cooperates with each other during the audit process to jointly complete the contract audit task.

[0077] This embodiment of the present invention abandons the traditional model of manual, word-by-word review and achieves rapid contract processing through automated processes. By rapidly configuring requirements in the front-end configuration module, intelligent agents in the back-end processing module work in parallel, significantly reducing review time. Manual review cycles that used to take hours or even days are compressed to minutes or even less, significantly accelerating business processes.

[0078] This embodiment of the present invention utilizes multi-agent collaboration and precise intent recognition to enable contract review across the full range of legal, financial, and business domains. Based on professional training and a rule library, these agents reduce human oversight and subjective judgment errors, accurately identifying risky clauses in contracts and helping businesses mitigate potential risks.

[0079] By combining the flexible configuration of the front-end configuration module with the dynamic response of the back-end processing module, this embodiment of the present invention enables the system to easily handle contract audits of varying complexity across different industries. Whether it's a simple commodity sales contract or a complex agreement involving the interests of multiple parties, audit strategies can be customized on demand, meeting the diverse and personalized contract audit needs of enterprises and significantly enhancing the system's versatility and practicality.

[0080] This embodiment generates a contract review request based on the configured contract review description information and processes the contract review request. First, the contract review description information is semantically parsed to extract key review elements. Then, the intent recognition agent is called from the agent library to perform multi-level matching of key review elements and the review rule library. Based on the attributes of the review intent, the corresponding pre-trained agent is matched, searched and called from the agent library. The called agent is used to review the contract content. The collaborative scheduling of multiple agents in the agent library is used to achieve automated and accurate review of the contract content.

[0081] like Figure 2 As shown, the contract automation review method based on multi-agent collaboration provided by the embodiment of the present invention is executed in the back-end processing module and includes the following steps:

[0082] S201: receiving a contract review request generated by configuring the contract review description information by the front-end configuration module, responding to the contract review request and performing semantic analysis on the contract review description information to extract key review elements.

[0083] The process of semantically parsing the contract review description information here is an existing technology and can adopt existing semantic models, such as the BERT model or the GPT model.

[0084] The extracted key audit elements include but are not limited to the audit objects and audit standards.

[0085] S202: Call the pre-trained intention recognition agent in the agent library, and use the intention recognition agent to perform multi-level matching of keywords, semantics and logic on the key audit elements and the pre-built audit rule library in turn to obtain the audit intent.

[0086] It should be noted that the agent library pre-stores an intent recognition agent, which is pre-trained using an existing large model. In this embodiment, the audit rule library has a dynamic update mechanism that can promptly update and improve audit rules based on changes in laws and regulations, adjustments to industry standards, and changes in the business needs of enterprises, ensuring the timeliness and accuracy of the rule library.

[0087] The process of obtaining the audit intent is as follows:

[0088] Step 2021: Match the key audit elements with the keywords in the pre-built audit rule library to preliminarily locate the scope of the rule entries in the audit rule library.

[0089] It should be noted here that the expression of key audit elements is consistent with the keyword expression in the audit rule library, which can improve the accuracy of matching.

[0090] Step 2022: Perform semantic identification on the key audit elements, analyze the contextual relationship of the key audit elements, determine whether they are consistent with the rule semantics within the scope of the initially located rule entries, and obtain the rule entries that are consistent with the rule semantics.

[0091] In the process of semantic matching, existing large models such as the BERT model are used for semantic understanding and recognition, to understand the semantic meaning of the audit elements, analyze their contextual relationships, and determine whether they are consistent with the semantics of the rules in the rule base.

[0092] Step 2023: Perform a logical relationship analysis on the key audit elements, and then, based on the results of the logical relationship analysis, select the final audit rules from the rule entries that match the rule semantics, and use the final audit rules as the audit intent.

[0093] In step 2023, the logical relationship includes but is not limited to conditional relationship, causal relationship, etc. For example, a decision tree model can be used to analyze the logical relationship of key audit elements, or existing methods such as the logical framework method can be used to analyze the logical relationship.

[0094] In one or more embodiments, after obtaining the audit intention, the method further includes:

[0095] Evaluate the audit intent to determine whether it is consistent with the contract audit description information, and revise the audit intent that is inconsistent with the contract audit description information.

[0096] If it is manually discovered that the audit intention is inconsistent with the overall content of the contract, business background, etc., or deviates from the audit requirements entered by the user, the audit intention can be manually judged and scored, thus forming a feedback mechanism. Users can correct their intentions by scoring the audit intention, tracing back the audit element parsing process, re-matching rules, etc.

[0097] At the same time, if new audit requirements or potential intentions that are not clearly expressed are discovered during the audit process, the intentions can be supplemented based on the audit rule base and relevant knowledge to ensure a comprehensive and accurate understanding of the contract audit requirements.

[0098] S203: Based on the attributes of the audit intention, match and search and call the corresponding pre-trained intelligent agent from the intelligent agent library, use the called intelligent agent to conduct a collaborative audit of the contract content to obtain the audit result.

[0099] In one or more embodiments, the audit results are evaluated to determine whether they meet established requirements. If the audit results meet established requirements, the audit is terminated. If the audit results do not meet established requirements, the audit intent is revised and supplemented until the audit results meet established requirements. This ensures the accuracy of the audit results.

[0100] In this embodiment, the audit intent attributes include auditing the legal attributes of the contract and auditing the financial attributes of the contract. When the audit intent attributes are auditing the legal attributes of the contract, a pre-trained legal agent is searched and invoked from the agent library; when the audit intent attributes are auditing the financial attributes of the contract, a pre-trained financial agent is searched and invoked from the agent library.

[0101] Each audit intent is associated with a fixed and unique agent ID in the agent library, allowing for the selection of a matching agent from multiple agent libraries to perform contract audits. Each agent is trained to possess audit capabilities specific to a specific domain. For example, a legal agent excels at reviewing legal risk clauses in contracts, while a financial agent accurately reviews financial data and clauses within contracts. The selected agent is then called upon to review the contract content and obtain the audit results.

[0102] This embodiment generates a contract review request based on the configured contract review description information and processes the contract review request. First, the contract review description information is semantically parsed to extract key review elements. Then, the intent recognition agent is called from the agent library to perform multi-level matching of key review elements and the review rule library. Based on the attributes of the review intent, the corresponding pre-trained agent is matched, searched and called from the agent library. The called agent is used to review the contract content. The collaborative scheduling of multiple agents in the agent library is used to achieve automated and accurate review of the contract content.

[0103] The structure of the electronic device according to the embodiment of the present invention is described in detail below. Figure 3 The schematic diagram of the structure of the electronic device provided in the embodiment of the present invention can be understood as follows: Figure 3 Only exemplary structures of the electronic device are shown, not all structures. Part or all of the shown structures may be implemented as needed.

[0104] The electronic device provided in the embodiment of the present invention includes: at least one processor 301, a memory 302, a user interface 303 and at least one network interface 304. The various components in the contract automation review system based on multi-agent collaboration are coupled together through a bus system 305. It can be understood that the bus system 305 is used to realize the connection and communication between these components. In addition to the data bus, the bus system 305 also includes a power bus, a control bus and a status signal bus. However, for the sake of clarity, Figure 3 Various buses are labeled as bus system 305 .

[0105] The user interface 303 may include a display, a keyboard, a mouse, a trackball, a click wheel, keys, buttons, a touch pad or a touch screen.

[0106] It will be appreciated that the memory 302 may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memory. The memory 302 in the embodiment of the present invention can store data to support the operation of the terminal. Examples of such data include any computer program used to operate on the terminal, such as an operating system and application programs. The operating system includes various system programs, such as a framework layer, a core library layer, and a driver layer, which are used to implement various basic services and handle hardware-based tasks. The application program may include various application programs.

[0107] In some embodiments, the automated contract review system based on multi-agent collaboration provided by the embodiments of the present invention is implemented using a combination of software and hardware. As an example, the automated contract review system based on multi-agent collaboration provided by the embodiments of the present invention may be a processor in the form of a hardware decoding processor, which is programmed to execute the automated contract review method based on multi-agent collaboration provided by the embodiments of the present invention. For example, the processor in the form of a hardware decoding processor may be one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic components.

[0108] As an example, the processor 301 can be an integrated circuit chip with signal processing capabilities, such as a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., where the general-purpose processor can be a microprocessor or any conventional processor, etc.

[0109] As an example of hardware implementation of the contract automation review system based on multi-agent collaboration provided by an embodiment of the present invention, the device provided by an embodiment of the present invention can be directly executed by a processor 301 in the form of a hardware decoding processor. For example, it can be executed by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs) or other electronic components to implement the contract automation review method based on multi-agent collaboration provided by an embodiment of the present invention.

[0110] The memory 302 in the embodiment of the present invention is used to store various types of data to support the operation of the contract automation review system based on multi-agent collaboration, or to store data for executing Figure 2 Examples of such data include any executable instructions for operating on a multi-agent collaborative contract automation review system, such as executable instructions. A program implementing the multi-agent collaborative contract automation review method according to an embodiment of the present invention may be included in the executable instructions.

[0111] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer readable medium, the computer program including a computer program for executing Figure 1 In such an embodiment, the computer program can be downloaded and installed from a network via the communication portion and / or installed from a removable medium. When the computer program is executed by the central processing unit, the various functions defined in the apparatus of the present application are performed.

[0112] The present invention is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products of the embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams and the combination of processes and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0113] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A contract automation review system based on multi-agent collaboration, characterized by: include: The front-end configuration module is used to: configure contract review description information and generate contract review requests; Backend processing module, which is used to: Receive a contract review request, respond to the contract review request, and perform semantic analysis on the contract review description information to extract key review elements; Call the pre-trained intent recognition agent in the agent library, and use the intent recognition agent to match the key audit elements with the pre-built audit rule library in keywords, semantics and logic at multiple levels to obtain the audit intent; the audit rule library has a dynamic update mechanism based on changes in laws and regulations, adjustments to industry standards and changes in the business needs of enterprises; Based on the attributes of the audit intent, the corresponding pre-trained agent is matched and called from the agent library, and the called agent is used to collaboratively review the contract content to obtain the audit results; In the backend processing module, the process of obtaining the audit intention is as follows: Match key audit elements with keywords in the pre-built audit rule library to preliminarily locate the scope of rule entries in the audit rule library; Perform semantic identification on key audit elements, analyze the contextual relationship of key audit elements, determine whether they are consistent with the semantics of rules within the scope of the initially located rule entries, and obtain rule entries with consistent semantics; Conduct a logical relationship analysis on key audit elements, and then, based on the results of the logical relationship analysis, select the final audit rules from the rule entries that match the rule semantics, and use the final audit rules as the audit intent; In the backend processing module, after obtaining the audit intention, the following steps are also included: Evaluate the audit intent to determine whether it is consistent with the contract audit description information, and revise the audit intent that is inconsistent with the contract audit description information; If the audit intent is found to be inconsistent with the overall contract content or business background, or deviates from the audit requirements input by the user, the audit intent will be manually judged and scored, forming a feedback mechanism to retrace the audit element parsing process and re-match the rules to correct the intent. At the same time, if new audit requirements or potential intentions that are not clearly expressed are discovered during the audit process, the intent will be supplemented based on the audit rule library and relevant knowledge to ensure a comprehensive and accurate understanding of the contract audit requirements. The back-end processing module also includes evaluating the audit results to determine whether the audit results meet the set requirements; when the audit results meet the set requirements, the audit is ended; when the audit results do not meet the set requirements, the audit intention is corrected and supplemented until the audit results meet the set requirements.

2. The automated contract review system based on multi-agent collaboration according to claim 1, characterized in that: In the back-end processing module, the attributes of the audit intention include the legal attributes in the audit contract and the financial attributes in the audit contract.

3. The automated contract review system based on multi-agent collaboration according to claim 1, characterized in that: In the back-end processing module, when the attribute of the audit intention is to audit the legal attribute in the contract, the pre-trained legal agent is matched, searched and called from the agent library; when the attribute of the audit intention is to audit the financial attribute in the contract, the pre-trained financial agent is matched, searched and called from the agent library.

4. The multi-agent collaborative contract automation review system according to claim 1, characterized in that: The multi-agent collaborative contract automation review system further includes: The audit result feedback module is used to feed back the audit results to the front-end configuration module.

5. A contract automation review method based on multi-agent collaboration, characterized in that: It is executed in the backend processing module and includes the following steps: Receive a contract review request generated by configuring the contract review description information by the front-end configuration module, respond to the contract review request and perform semantic analysis on the contract review description information to extract key review elements; Call the pre-trained intent recognition agent in the agent library, and use the intent recognition agent to match the key audit elements with the pre-built audit rule library in keywords, semantics and logic at multiple levels to obtain the audit intent; the audit rule library has a dynamic update mechanism based on changes in laws and regulations, adjustments to industry standards and changes in the business needs of enterprises; Based on the attributes of the audit intent, the corresponding pre-trained agent is matched and called from the agent library, and the called agent is used to collaboratively review the contract content to obtain the audit results; After obtaining the audit intent, it also includes: Evaluate the audit intent to determine whether it is consistent with the contract audit description information, and revise the audit intent that is inconsistent with the contract audit description information; If the audit intent is found to be inconsistent with the overall contract content or business background, or deviates from the audit requirements input by the user, the audit intent will be manually judged and scored, forming a feedback mechanism to retrace the audit element parsing process and re-match the rules to correct the intent. At the same time, if new audit requirements or potential intentions that are not clearly expressed are discovered during the audit process, the intent will be supplemented based on the audit rule library and relevant knowledge to ensure a comprehensive and accurate understanding of the contract audit requirements. The automated contract audit method based on multi-agent collaboration also includes evaluating the audit results to determine whether the audit results meet the set requirements; when the audit results meet the set requirements, the audit is ended; when the audit results do not meet the set requirements, the audit intention is corrected and supplemented until the audit results meet the set requirements.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps in the contract automation review method based on multi-agent collaboration as described in claim 5 are implemented.

Images