CN115269874A - Intelligent contract examination method based on natural language understanding - Google Patents
Intelligent contract examination method based on natural language understanding Download PDFInfo
- Publication number
- CN115269874A CN115269874A CN202210917535.XA CN202210917535A CN115269874A CN 115269874 A CN115269874 A CN 115269874A CN 202210917535 A CN202210917535 A CN 202210917535A CN 115269874 A CN115269874 A CN 115269874A
- Authority
- CN
- China
- Prior art keywords
- examination
- contract
- points
- information
- label
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/18—Legal services; Handling legal documents
Abstract
The invention relates to the technical field of document examination, and discloses a contract intelligent examination method based on natural language understanding, which comprises the following steps: s1, constructing a contract examination knowledge graph: manually constructed label systems and examination points jointly form a contract examination knowledge graph; s2, implementing the examination rule; the algorithm engineer realizes a specific contract examination function according to the knowledge graph; and S3, intelligently checking the contract. The intelligent contract examination method based on natural language understanding combs contract examination requirements, forms a contract examination knowledge graph containing labels, examination points, examination items, examination rules, risk prompt information and the like, trains an algorithm model according to the contract examination knowledge graph, realizes the examination rules, can generate an examination list according to the content of an indication graph, intelligently dispatches a related algorithm model and an examination module according to the difference of the examination list and the examination standpoint, and finally obtains a contract examination result.
Description
Technical Field
The invention belongs to the technical field of document examination, and particularly relates to an intelligent contract examination method based on natural language understanding.
Background
The contract is a contract carrier of business cooperation between enterprises and is used as indispensable work content of daily operation of an enterprise main body, the traditional contract examination method greatly depends on manual examination operation and occupies a large amount of corporate legal human resources for a long time, meanwhile, because the experience of each contract examiner is different, the law and regulation are understood differently, the risk points are controlled differently, and the final contract examination standard and result are different, the compliance of enterprise operation and the risk of dispute litigation are increased invisibly.
The intelligent examination of the contract, as a new functional service, appears in some paperless office services or document management services, but the examination of the contract mainly compares the current contract with the template contract, and judges whether risks exist according to the difference of texts, so that the examination of the contract is very mechanical, has no obvious effect on reducing the workload of legal affairs, even adds some steps of examination of the contract, and cannot really solve the problems of complicated work, non-uniform standards, manpower consumption and the like in the examination of the contract.
Disclosure of Invention
In view of the above situation, in order to overcome the defects in the prior art, the present invention provides an intelligent contract review method based on natural language understanding, which effectively solves the problems presented in the background art.
In order to achieve the purpose, the invention provides the following technical scheme: an intelligent contract examination method based on natural language understanding, comprising the following steps:
s1, constructing a contract examination knowledge graph: manually constructed label systems and examination points jointly form a contract examination knowledge graph;
s2, realizing the examination rule: the algorithm engineer realizes a specific contract examination function according to the knowledge graph;
s3, intelligent examination of contracts: before contract examination, a user can customize an examination list, all examination points of each contract type are combined into a default examination list of the contract type, the user can screen out the examination points required by the user from the default examination list, the examination point names, the sub-examination item names and the risk prompt information of the examination points can be edited, the examination list required by the user is finally formed, and different examination operations can be automatically executed according to the difference of the examination points related to each examination list through the subsequent intelligent examination function.
Preferably, the constructing of the contract inspection knowledge-graph in S1 includes the following steps:
s1-1, combing the contract examination requirements, and integrating the relatively abstract contract examination requirements into a plurality of limited examination points, for example, the contract money examination points in the goods buying and selling contracts;
s1-2, integrating various risks contained in the examination points into a plurality of limited sub examination items, wherein each examination item examines a specific risk point, and for example, under the examination points of contract amount examination, a plurality of sub examination items such as 'consistency of capital and small form', 'total amount accuracy' and the like are available;
s1-3, abstracting a label possibly required for realizing the examination logic according to the examination point and the sub examination items thereof, determining the content of the label, randomly selecting some real contracts, tentatively labeling the label, and confirming whether the label is available and the content is accurate in the real contracts;
s1-4, if in the process of test annotation, certain labels are unreasonably arranged or the connotation is inaccurate, repeating the two steps, iterating the labels and the connotation to the extent that an algorithm is available and the connotation is clear, integrating the labels into a label system with a certain hierarchical relationship, for example, the goods buying and selling contract has 'payment' terms and 'default liability' terms, and the 'payment' terms have 'total price of goods', 'delivery time' and other terms, and sorting the terms into a group of labels with a hierarchical structure;
s1-5, labeling a series of contract text data according to the constructed label system and label connotation, and synchronizing the label system and the labeled data to an algorithm engineer for training an algorithm model;
s1-6, forming a specific examination rule of each sub examination item of the examination point according to a label system, such as ' total amount of contract in total price of goods ' label, and if the total amount of capital, the total amount of lowercase and the currency type do not exist at the same time, the risk exists ';
s1-7, forming a contract examination knowledge graph by all examination points, wherein the graph comprises all labels, the examination points, examination items, examination rules, risk prompt information and the like;
and S1-8, synchronizing the combed knowledge graph to an algorithm engineer by using a contract knowledge graph synchronization module to realize examination points and examination rules, and simultaneously adjusting knowledge graph information such as related labels/examination rules and the like according to the intelligent examination actual effect of the contract to optimize the iterative contract examination knowledge graph.
Preferably, the implementing the examination rule in S2 includes the following steps:
s2-1, extracting relevant information of a label system by using a contract knowledge graph synchronization module, and generating training data of a model according to a label hierarchical structure by using the labeled data based on the label system;
s2-2, training an algorithm model based on the label system and the training data, so that the algorithm can accurately identify various labels and elements in the contract according to the artificially constructed label system;
s2-3, extracting examination points, sub-examination items, examination rules and related label information by using a contract knowledge map synchronization module, realizing the examination rules of all the sub-examination items of each examination point by using model labels and elements, packaging the examination points into an examination module which can be called, for example, in the examination points of 'total contract amount examination', obtaining a section of terms of the total contract amount by using a label model, then extracting specific total amount elements in the section of terms by using an element extraction model, further realizing subsequent examination rules by using the total amount elements, and packaging the examination rules into the examination module, so that the examination points can be called at any time and examine related risks of 'total contract amount';
s2-4, testing the effect of the intelligent contract inspection algorithm by using the real contract, if the inspection result is wrong, finely adjusting the inspection rule under the wrong inspection item, synchronizing the fine adjustment result to the contract inspection knowledge map, and finally iterating the intelligent contract inspection algorithm with the best effect;
and S2-5, extracting examination points, sub-examination items, risk prompt texts and related label information by using a contract knowledge map synchronization module, and generating an examination list.
Preferably, the intelligent examination of the contract in S3 includes the following steps:
s3-1, uploading the contract by the user, starting to inspect, and analyzing the contract in the formats of docx, PDF and the like into contract text information in a unified format by the document analysis module;
s3-2, the element extraction module can firstly identify basic information in the contract, such as contract types and information of all parties signing the contract, the information is displayed to a user, necessary information for completing intelligent examination is assisted to the user, and corresponding examination lists are selected, for example, examination points of a 'shop lease' type contract and a 'goods buy and sell' type contract are different, in the 'goods buy and sell' type contract, examination key points concerned by whether the user belongs to a 'buyer' or a 'seller' are also different, and the element extraction module can assist the user to quickly select necessary information in the contract examination and select the examination lists so as to continue a subsequent examination step;
s3-3, the user autonomously selects an examination place and an examination point list, a subsequent intelligent examination step is continued, then an algorithm scheduling module screens examination modules needing to be executed according to the examination place and the examination place, and the modules are combined into an examination module execution tree needing to be executed in the examination; acquiring all clause classification models and element extraction models required by the examination modules, generating an algorithm model execution tree required to be executed in the examination according to the hierarchical relation between all labels and elements described by a label system, and sequentially calling all models and examination modules according to the algorithm model execution tree and the examination module execution tree;
s3-4, an examination module, which is used for acquiring clause information and element information required by examination from the execution results of the clause classification model and the element extraction model, sequentially executing all sub examination items according to the information in the contract examination knowledge map, judging whether risks exist according to examination rules of the sub examination items, and generating and returning all examination results;
s3-5, generating a risk card of specific risk prompt information according to the result returned by the examination module and the configuration of an examination list, and then enabling a user to choose to ignore the risk or annotate related risks according to the prompt information of the risk card, wherein the risk information and annotation information which are not ignored are stored on a contract examination platform, and optionally exporting a contract, and the exported contract can also store the risk prompt information and the annotation information of the user at the same time in an annotation mode;
s3-6, after the user modifies the contract on the contract examination platform or uploads the modified contract again, the modified contract is stored as a new version of the original contract, the contract of each version can be compared with the contract of the original contract by using a version comparison module to confirm the modified content of each version of the contract, and the new version of the contract can also be intelligently examined to confirm the risk condition;
and S3-7, after all risk information of the confirmed contract is released or all risks are ignored, a contract signing process can be started.
Compared with the prior art, the invention has the beneficial effects that:
1. legal knowledge involved in the process of combing contract examination is solidified, a contract examination knowledge map is solidified, a contract examination method is standardized, an algorithm model is trained to understand contract contents and examination rules are realized, so that intelligent examination of contracts is realized, and efficiency of contract examination and normalization of contract examination are further improved. Meanwhile, special customization of the examination requirements of a specific contract can be realized, and the thinking flow and the examination flow of legal personnel during examination and contract are restored to the maximum extent by using a machine.
2. According to different examination requirements, a user can select different examination points, examination lists, sub-examination items and contract holding parties, different examination rules are executed according to different examination scenes, and the user can edit the names of the examination points, the risk prompt information and other contents by himself, so that the intelligent contract examination is closer to the real examination requirements of the user.
3. The inspection algorithm realizes the processes and the contract inspection knowledge map construction process, the two processes iteratively evolve in the processes to improve the effect, and meanwhile, the output result can assist the optimization iteration of the other process, so that the two processes can finally achieve the best effect more efficiently.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention. In the drawings:
FIG. 1 is a schematic flow diagram of the present invention;
FIG. 2 is a schematic diagram of an intelligent contract review process according to the present invention;
FIG. 3 is a flow chart illustrating the implementation of the examination rule according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments; all other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.
In a first embodiment, as shown in fig. 1, the present invention includes an intelligent examination method for contracts based on natural language understanding, where the examination method includes the following steps:
s1, constructing a contract examination knowledge graph: manually constructed label systems and examination points jointly form a contract examination knowledge graph;
s2, realizing the examination rule: the algorithm engineer realizes a specific contract examination function according to the knowledge graph;
s3, intelligent examination of contracts: before contract examination, a user can customize an examination list, all examination points of each contract type are combined into a default examination list of the contract type, the user can screen out the examination points required by the user from the default examination list, the examination point names, the sub-examination item names and the risk prompt information of the examination points can be edited, the examination list required by the user is finally formed, and different examination operations can be automatically executed according to the difference of the examination points related to each examination list through the subsequent intelligent examination function.
In the second embodiment, the step of constructing the contract inspection knowledge-graph in S1 comprises the following steps:
s1-1, combing the contract examination requirements, and integrating the relatively abstract contract examination requirements into a plurality of limited examination points, for example, the contract money examination points in the goods buying and selling contracts;
s1-2, integrating various risks contained in the examination points into a plurality of limited sub examination items, wherein each examination item examines a specific risk point, and for example, under the examination points of contract amount examination, a plurality of sub examination items such as 'consistency of capital and small form', 'total amount accuracy' and the like are available;
s1-3, abstracting a label possibly required for realizing the examination logic according to the examination point and the sub examination items thereof, determining the content of the label, randomly selecting some real contracts, tentatively labeling the label, and confirming whether the label is available and the content is accurate in the real contracts;
s1-4, if in the process of test annotation, certain labels are unreasonably arranged or the connotation is inaccurate, repeating the two steps, iterating the labels and the connotation to the extent that an algorithm is available and the connotation is clear, integrating the labels into a label system with a certain hierarchical relationship, for example, the goods buying and selling contract has 'payment' terms and 'default liability' terms, and the 'payment' terms have 'total price of goods', 'delivery time' and other terms, and sorting the terms into a group of labels with a hierarchical structure;
s1-5, labeling a series of contract text data according to the constructed label system and label connotation, and synchronizing the label system and the labeled data to an algorithm engineer for training an algorithm model;
s1-6, forming a specific examination rule of each sub examination item of the examination point according to a label system, wherein the specific examination rule comprises a 'total price of goods' and a contract total amount in a label, and if a capital amount, a lowercase amount and a currency type do not exist at the same time, a risk exists;
s1-7, forming a contract examination knowledge map by all examination points, wherein the map comprises all labels, the examination points, examination items, examination rules, risk prompt information and the like;
and S1-8, synchronizing the combed knowledge graph to an algorithm engineer by using a contract knowledge graph synchronization module to realize examination points and examination rules, and simultaneously adjusting knowledge graph information such as related labels/examination rules and the like according to the intelligent examination actual effect of the contract to optimize the iterative contract examination knowledge graph.
In a third embodiment, the implementation of the examination rule in S2 includes the following steps:
s2-1, extracting relevant information of a label system by using a contract knowledge graph synchronization module, and generating training data of a model according to a label hierarchical structure by using the labeled data based on the label system;
s2-2, training an algorithm model based on the label system and training data, so that the algorithm can accurately identify various labels and elements in the contract according to the artificially constructed label system;
s2-3, extracting examination points, sub-examination items, examination rules and related label information by using a contract knowledge map synchronization module, realizing the examination rules of all the sub-examination items of each examination point by using model labels and elements, packaging the examination points into an examination module which can be called, for example, in the examination points of 'total amount of contracts' examination, obtaining the clause of the total amount of the contracts by using a label model, then extracting specific total amount elements in the clause by using an element extraction model, further realizing subsequent examination rules by using the total amount elements, and after packaging the examination rules into the examination module, calling the examination points at any time and examining related risks of 'total amount of contracts';
s2-4, testing the effect of the intelligent contract inspection algorithm by using the real contract, if the inspection result is wrong, finely adjusting the inspection rule under the wrong inspection item, synchronizing the fine adjustment result to the contract inspection knowledge map, and finally iterating the intelligent contract inspection algorithm with the best effect;
and S2-5, extracting examination points, sub-examination items, risk prompt texts and related label information by using a contract knowledge map synchronization module, and generating an examination list.
In a fourth embodiment, the intelligent examination of the contract in S3 includes the following steps:
s3-1, uploading the contract by the user, starting to inspect, and analyzing the contract in the formats of docx, PDF and the like into contract text information in a unified format by the document analysis module;
s3-2, the element extraction module can firstly identify basic information in the contract, such as contract types and information of all parties signing the contract, the information is displayed to a user, necessary information for completing intelligent examination is assisted to the user, and corresponding examination lists are selected, for example, examination points of a 'shop lease' type contract and a 'goods buy and sell' type contract are different, in the 'goods buy and sell' type contract, examination key points concerned by whether the user belongs to a 'buyer' or a 'seller' are also different, and the element extraction module can assist the user to quickly select necessary information in the contract examination and select the examination lists so as to continue a subsequent examination step;
s3-3, the user autonomously selects an examination place and an examination point list, a subsequent intelligent examination step is continued, then an algorithm scheduling module screens examination modules needing to be executed according to the examination place and the examination place, and the modules are combined into an examination module execution tree needing to be executed in the examination; acquiring all clause classification models and element extraction models required by the examination modules, generating an algorithm model execution tree required to be executed in the examination according to the hierarchical relation between all labels and elements described by a label system, and sequentially calling all models and examination modules according to the algorithm model execution tree and the examination module execution tree;
s3-4, an examination module, which is used for acquiring clause information and element information required by examination from execution results of the clause classification model and the element extraction model, sequentially executing all sub-examination items according to information in the contract examination knowledge graph, judging whether risks exist according to examination rules of the sub-examination items, and generating and returning all examination results;
s3-5, generating a risk card of specific risk prompt information according to the result returned by the examination module and the configuration of an examination list, and then enabling a user to choose to ignore the risk or annotate the related risk according to the prompt information of the risk card, wherein the risk information and the annotation information which are not ignored are stored on a contract examination platform, and optionally exporting a contract, and the exported contract can also store the risk prompt information and the user annotation information in an annotation mode;
s3-6, after the user modifies the contract on the contract examination platform or uploads the modified contract again, the modified contract is stored as a new version of the original contract, the contract of each version can be compared with the contract of the original contract by using a version comparison module to confirm the modified content of each version of the contract, and the new version of the contract can also be intelligently examined to confirm the risk condition;
and S3-7, after all risk information of the confirmed contract is released or all risks are ignored, a contract signing process can be started.
The label indicates that a piece of contract text has special meanings, such as 'default responsibility', 'confidentiality clause', 'payment clause' and the like;
the above elements represent some key segment information in a section of contract text, such as 'contract main body', 'contract total amount', etc.;
the label system refers to the relationship among the labels, the label and the element, and the element, for example, the label of "payment clause" and the label of "default liability" are in parallel, the label of "payment clause" is provided with the label of "total price of goods" and the label of "delivery time", and the label of "total price of goods" is provided with the element of "total amount of contract", etc.;
the examination point refers to a specific examination point in the contract examination, such as "examination of dispute resolution terms" and the like, and usually comprises several examination items;
the examination item refers to an examination point in the examination point, for example, a "dispute resolution term missing risk" examination item, a "simultaneous engagement litigation and arbitration risk" examination item, and the like in the "dispute resolution term examination" examination point, and usually a label, an element, and a certain examination rule are required to be written for one examination item;
the examination rules refer to judgment rules of certain risks in contract examination, such as that the default fund is risky when being more than 30% of the total amount;
the examination result refers to the final execution result of the examination item, and comprises a result name, whether the examination item is risky, risk prompt information and the like;
the examination module is used for executing specific examination points;
the examination list refers to a set of all examination points aiming at a certain contract type and is used for examination of a specific contract;
the examination places refer to examination places of users, such as lessees, buyers and sellers, which are different in examination places and examination rules, and examination points, examination items and examination rules are different when examination is closed;
the contract examination knowledge graph is a knowledge graph which is combed according to a certain type of contract and contains all labels, elements, examination points and examination rules;
the term classification model refers to an algorithm model which identifies key paragraphs in the contract text and marks corresponding labels, for example, identifies the "default liability" terms in the contract and marks the "default liability" labels;
the element extraction model refers to an algorithm model for identifying the key segments in the paragraphs and marking corresponding labels, for example, identifying the total amount of the goods in the payment terms and marking the label of the total amount of the goods;
the legal researcher refers to talents with legal background and professional knowledge and is responsible for defining labels and elements, carding a label system, examining and checking points, examining items and examining rules;
the algorithm engineer refers to an algorithm engineer and is responsible for realizing a clause classification model, a factor extraction model and a review project.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (4)
1. An intelligent contract examination method based on natural language understanding is characterized in that: the examination method comprises the following steps:
s1, constructing a contract examination knowledge graph: manually constructed label systems and examination points jointly form a contract examination knowledge graph;
s2, implementing the examination rule: an algorithm engineer realizes a specific contract examination function according to the knowledge graph;
s3, intelligent examination of contracts: before contract examination, a user can customize an examination list, all examination points of each contract type are combined into a default examination list of the contract type, the user can screen out the examination points required by the user from the default examination list, the examination point names, the sub-examination item names and the risk prompt information of the examination points can be edited, the examination list required by the user is finally formed, and different examination operations can be automatically executed according to the difference of the examination points related to each examination list through the subsequent intelligent examination function.
2. The intelligent examination method for contracts based on natural language understanding of claim 1, wherein: the method for constructing the contract examination knowledge graph in the S1 comprises the following steps:
s1-1, combing the contract examination requirements, and integrating the relatively abstract contract examination requirements into a plurality of limited examination points, for example, the contract money examination points in the goods buying and selling contracts;
s1-2, integrating various risks contained in the examination points into a plurality of limited sub examination items, wherein each examination item examines a specific risk point, for example, the examination points of contract amount examination have a plurality of sub examination items such as ' consistency of capital and small capital, and ' total amount accuracy ';
s1-3, abstracting a label possibly required for realizing the examination logic according to the examination point and the sub examination items thereof, determining the content of the label, randomly selecting some real contracts, tentatively labeling the label, and confirming whether the label is available and the content is accurate in the real contracts;
s1-4, if in the process of test annotation, certain labels are unreasonably arranged or the connotation is inaccurate, repeating the two steps, iterating the labels and the connotation to the extent that an algorithm is available and the connotation is clear, integrating the labels into a label system with a certain hierarchical relationship, for example, the goods buying and selling contract has 'payment' terms and 'default liability' terms, and the 'payment' terms have 'total price of goods', 'delivery time' and other terms, and sorting the terms into a group of labels with a hierarchical structure;
s1-5, labeling a series of contract text data according to the constructed label system and label connotation, and synchronizing the label system and the labeled data to an algorithm engineer for training an algorithm model;
s1-6, forming a specific examination rule of each sub examination item of the examination point according to a label system, such as ' total amount of contract in total price of goods ' label, and if the total amount of capital, the total amount of lowercase and the currency type do not exist at the same time, the risk exists ';
s1-7, forming a contract examination knowledge graph by all examination points, wherein the graph comprises all labels, the examination points, examination items, examination rules, risk prompt information and the like;
and S1-8, synchronizing the combed knowledge graph to an algorithm engineer by using a contract knowledge graph synchronization module to realize examination points and examination rules, and simultaneously adjusting knowledge graph information such as related labels/examination rules and the like according to the intelligent examination actual effect of the contract to optimize the iterative contract examination knowledge graph.
3. The intelligent examination method for contracts based on natural language understanding of claim 1, characterized in that: the implementation of the examination rule in the step S2 includes the following steps:
s2-1, extracting relevant information of a label system by using a contract knowledge graph synchronization module, and generating training data of a model according to a label hierarchical structure by using the labeled data based on the label system;
s2-2, training an algorithm model based on the label system and training data, so that the algorithm can accurately identify various labels and elements in the contract according to the artificially constructed label system;
s2-3, extracting examination points, sub-examination items, examination rules and related label information by using a contract knowledge map synchronization module, realizing the examination rules of all the sub-examination items of each examination point by using model labels and elements, packaging the examination points into an examination module which can be called, for example, in the examination points of 'total amount of contracts' examination, obtaining the clause of the total amount of the contracts by using a label model, then extracting specific total amount elements in the clause by using an element extraction model, further realizing subsequent examination rules by using the total amount elements, and after packaging the examination rules into the examination module, calling the examination points at any time and examining related risks of 'total amount of contracts';
s2-4, testing the effect of the intelligent contract inspection algorithm by using the real contract, if the inspection result is wrong, finely adjusting the inspection rule under the wrong inspection item, synchronizing the fine adjustment result to the contract inspection knowledge map, and finally iterating the intelligent contract inspection algorithm with the best effect;
and S2-5, extracting examination points, sub-examination items, risk prompt texts and related label information by using a contract knowledge map synchronization module, and generating an examination list.
4. The intelligent examination method for contracts based on natural language understanding of claim 1, characterized in that: the intelligent examination of the contract in the S3 comprises the following steps:
s3-1, uploading the contract by a user, starting to inspect, and simultaneously analyzing the contract with the formats of docx, PDF and the like into contract text information with a unified format by a document analysis module;
s3-2, the element extraction module can firstly identify basic information in the contract, such as contract types and information of all parties signed up, the information is displayed to a user, the necessary information for completing intelligent examination by the user is assisted, and a corresponding examination list is selected, for example, examination points of a contract of 'shop leasing' type and a contract of 'goods buying and selling' type are different, in the contract of 'goods buying and selling' type, examination key points concerned by whether the user belongs to 'buying part' or 'selling part' are also different, and the element extraction module can assist the user to quickly select the necessary information in the contract examination and select the examination list so as to continue a subsequent examination step;
s3-3, a user autonomously selects an examination place and an examination point list, a subsequent intelligent examination step is continued, then an algorithm scheduling module screens examination modules needing to be executed according to the examination place and the examination place, and the modules are combined into an examination module execution tree needing to be executed in the examination; acquiring all clause classification models and element extraction models required by the examination modules, generating an algorithm model execution tree required to be executed by the examination according to the hierarchical relationship between all labels and elements described by a label system, and sequentially calling all models and the examination modules according to the algorithm model execution tree and the examination module execution tree;
s3-4, an examination module, which is used for acquiring clause information and element information required by examination from execution results of the clause classification model and the element extraction model, sequentially executing all sub-examination items according to information in the contract examination knowledge graph, judging whether risks exist according to examination rules of the sub-examination items, and generating and returning all examination results;
s3-5, generating a risk card of specific risk prompt information according to the result returned by the examination module and the configuration of an examination list, and then enabling a user to choose to ignore the risk or annotate related risks according to the prompt information of the risk card, wherein the risk information and annotation information which are not ignored are stored on a contract examination platform, and optionally exporting a contract, and the exported contract can also store the risk prompt information and the annotation information of the user at the same time in an annotation mode;
s3-6, after the user modifies the contract on the contract examination platform or the offline modified contract is uploaded again, the modified contract is stored as a new contract of the original contract, the contracts of all versions can be compared with the contract of the original contract by using a version comparison module to confirm the modified content of each version of the contract, and the new contract can also be intelligently examined to confirm the risk condition;
and S3-7, after all risk information of the confirmed contract is released or all risks are ignored, a contract signing process can be started.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210917535.XA CN115269874A (en) | 2022-08-01 | 2022-08-01 | Intelligent contract examination method based on natural language understanding |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210917535.XA CN115269874A (en) | 2022-08-01 | 2022-08-01 | Intelligent contract examination method based on natural language understanding |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115269874A true CN115269874A (en) | 2022-11-01 |
Family
ID=83746854
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210917535.XA Pending CN115269874A (en) | 2022-08-01 | 2022-08-01 | Intelligent contract examination method based on natural language understanding |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115269874A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20230059494A1 (en) * | 2021-08-19 | 2023-02-23 | Digital Asset Capital, Inc. | Semantic map generation from natural-language text documents |
| CN117132244A (en) * | 2023-10-26 | 2023-11-28 | 国网浙江省电力有限公司 | Classification processing method, device and storage medium for intelligent compliance management system |
-
2022
- 2022-08-01 CN CN202210917535.XA patent/CN115269874A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20230059494A1 (en) * | 2021-08-19 | 2023-02-23 | Digital Asset Capital, Inc. | Semantic map generation from natural-language text documents |
| US20230056987A1 (en) * | 2021-08-19 | 2023-02-23 | Digital Asset Capital, Inc. | Semantic map generation using hierarchical clause structure |
| CN117132244A (en) * | 2023-10-26 | 2023-11-28 | 国网浙江省电力有限公司 | Classification processing method, device and storage medium for intelligent compliance management system |
| CN117132244B (en) * | 2023-10-26 | 2024-01-09 | 国网浙江省电力有限公司 | Classification processing method, device and storage medium for intelligent compliance management system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115269874A (en) | Intelligent contract examination method based on natural language understanding | |
| Nissen et al. | Integrated analysis and design of knowledge systems and processes | |
| Dalmaris et al. | A framework for the improvement of knowledge‐intensive business processes | |
| CN108564300B (en) | Efficient DWG drawing online collaborative verification method based on process engine | |
| Olsina et al. | Updating quality models for evaluating new generation web applications | |
| CN111861366A (en) | Project-ground intelligent auditing system and computer | |
| CN110378206B (en) | Intelligent image examination system and method | |
| Johannsen et al. | Meta modeling for business process improvement | |
| US8234570B2 (en) | Harvesting assets for packaged software application configuration | |
| CN115983571A (en) | Construction project auditing method and system based on artificial intelligence for construction industry | |
| TW201810172A (en) | System and method of standard-compliant electronic case report form design and clinical data set generation | |
| CN110533379A (en) | A kind of double record scene copy generating systems of insurance Internet-based | |
| Reed | Methodology for real time systems | |
| CN111400529B (en) | Data processing method and device | |
| Jeong et al. | Requirements development process for manufacturing information systems based on quality function deployment | |
| Carrillo et al. | Commonalities and differences between requirements engineering tools: A quantitative approach | |
| Thom et al. | Workflow patterns for business process modeling | |
| CN115841365A (en) | Model selection and quotation method, system, equipment and medium based on natural language processing | |
| JP2012088932A (en) | Accounting system, accounting method and program | |
| CN114035783A (en) | Software code knowledge graph construction method and tool | |
| Nardello et al. | Process model automation for industry 4.0: Challenges for automated model generation based on laboratory experiments | |
| Wajid et al. | Designing and evaluating a system of document recognition to support interoperability among collaborative enterprises | |
| CN215642785U (en) | Electronic inspection system for clinical test | |
| Germonprez et al. | The routinization of open source project engagement: The case of open source risk management routines | |
| Schön | The influence of automated planning on the task performance of process modelers |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |