Contract comprehensive risk prediction method and system and computer equipment
By acquiring the clause weights of contract text data and the capability and credit weights of historical cooperation data, comprehensive risk data is calculated, and risk modification items and prompts are generated. This solves the problem of inaccurate contract risk prediction in existing technologies and realizes the feasibility of multi-dimensional risk assessment and modification suggestions.
Patent Information
- Application Number
- CN202411118442.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-15
- Publication Date
- 2024-11-15
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing automated contract review methods only review the text content and do not consider other risk factors, resulting in low accuracy and feasibility of contract risk prediction results.
By acquiring the clause weights of contract text data and combining them with historical cooperation data to determine the capability and credit weight of the target partner, comprehensive risk data is calculated. Based on preset risk thresholds, risk modification items and prompts are generated to conduct multi-dimensional risk assessment.
It improves the accuracy of contract risk prediction and the feasibility of modification suggestions, avoids contract risks caused by word traps and insufficient performance capabilities of partners, and enhances the acceptance of the contract by both parties.
Smart Images

Figure CN118966778A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of data risk prediction and processing, and in particular to a method, system and computer equipment for comprehensive contract risk prediction. Background Art
[0002] Contracts are an extremely important part of all business activities. With the rapid development of modern business, the risks that may exist in contracts are gradually receiving attention. In order to avoid the risks that may exist in contracts, it is essential to review the contracts in order to predict the risks of the contracts.
[0003] In the prior art, the risk prediction of contracts is usually conducted by professionals in the field. However, manual review is not only costly but also inefficient. Therefore, automatic contract review methods are gradually gaining attention.
[0004] However, the existing automatic contract review method only performs a simple text content review on the existing contract text, and does not consider other risk factors, resulting in low accuracy of contract risk prediction results and low feasibility of risk recommendations. Summary of the invention
[0005] Based on this, the purpose of the present invention is to provide a method, system and computer device for predicting comprehensive contract risks, by first obtaining the clause weights in the contract text data, calculating the text risk data, and then judging the target partner's ability weight and credit weight based on historical cooperation data, and calculating the comprehensive risk data, so as to obtain risk modification items based on the text risk data and the comprehensive risk data, not only to judge the risk of text traps at the text level, but also to conduct multi-dimensional considerations on the partner's performance ability and overall credit to judge the risk clauses of the contract, while taking into account the strengths and weaknesses of the contract partners and the acceptance of negotiations. The present invention greatly improves the accuracy of contract risk prediction and the feasibility of modification suggestions.
[0006] The present invention proposes a method for predicting comprehensive contract risks, comprising: Acquire contract text data and perform word segmentation processing, perform classification and weighted processing on the contract text data after word segmentation processing to obtain clause weights, and obtain text risk data based on the clause weights; Obtain historical cooperation data to determine the capability weight and credit weight of the target partner, and calculate and obtain comprehensive risk data based on the capability weight and credit weight; The text risk data and the comprehensive risk data are judged according to a preset risk threshold to obtain risk modification items, so as to generate modification suggestions and final risk prompt items according to the risk modification items.
[0007] In summary, according to the above-mentioned comprehensive contract risk prediction method, by first obtaining the clause weights in the contract text data, calculating the text risk data, and then judging the target partner's ability weight and credit weight based on the historical cooperation data, and calculating the comprehensive risk data, the risk modification items are obtained based on the text risk data and the comprehensive risk data. Not only is the risk judgment of the text trap made at the text level, but also the multi-dimensional consideration of the partner's performance ability and overall credit is made to judge the risk clauses of the contract, while taking into account the strengths and weaknesses of the contract partners and the acceptance of the negotiation. The present invention greatly improves the accuracy of contract risk prediction and the feasibility of modification suggestions. The contract text data is obtained and segmented, and the contract text data after the segmentation is classified and weighted to obtain the clause weight, and the text risk data is obtained according to the clause weight, so as to avoid the cooperation risk that may be caused by text traps or disadvantageous clauses, and the historical cooperation data is obtained to judge the ability weight and credit weight of the target partner, and the comprehensive risk data is obtained according to the ability weight and the credit weight. The multi-dimensional considerations are made from the partners' performance ability and overall credit, so as to avoid the contract breach caused by the legal risk and financial risk of the partner, and the text risk data and the comprehensive risk data are judged according to the preset risk threshold to obtain the risk modification items, so as to generate the modification suggestions and the final risk warning items according to the risk modification items, and at the same time of multi-dimensional risk warnings, the clause modification suggestions are made according to the advantages and disadvantages of the ability and the negotiation position, and at the same time, the acceptance of the cooperation parties is further guaranteed. The present invention greatly improves the accuracy of contract risk prediction and the feasibility of modification suggestions.
[0008] Furthermore, the step of obtaining the contract text data and performing word segmentation processing specifically includes: Acquire contract text data and perform preprocessing, wherein the preprocessing includes text cleaning, text format normalization, and text detection and filtering; The text cleaning includes removing special characters, blank characters and stop words in the contract text data; The text format normalization includes respectively formatting the Chinese text, Western text, digital text and time text in the contract text data according to the preset contract text specification; The text detection and filtering includes detecting and filtering the template application paragraphs, standard statement paragraphs and regulatory reference paragraphs in the contract text data; Then, the pre-processed contract text data is subjected to text segmentation to obtain single sentence data, and the single sentence data is subjected to word segmentation processing, wherein the word segmentation processing includes word segmentation and sentence component labeling; The word segmentation includes identifying and determining whether a multi-character combination in the single sentence data is a single word; If it is determined that the multi-character combination is not a single word, the multi-character combination is divided into secondary characters according to the secondary segmentation method and a single word determination is performed; If it is determined that the multiple characters are combined into a single word, extracting the single word to combine into single word data; The sentence component tagging includes dividing the single sentence data into sentence components, obtaining sentence component character segments in the single sentence data, and matching the single word data with the sentence component character segments to tag the single word data with a sentence label, wherein the sentence label includes component position information of the single word data in the corresponding single sentence data and paragraph position information of the single sentence data in the contract text data; The single word data after word segmentation processing is input into the classification weighting module.
[0009] Furthermore, after the step of inputting the single word data after word segmentation processing into the classification weighting module, it also includes: Perform keyword matching on single word data according to a preset keyword database to obtain key word data, and locate and obtain key sentence data according to paragraph position information in sentence tags; Classifying the key sentence data into clause types according to the keyword types to obtain multiple types of key sentence data; Then, all the multi-type key sentence data are labeled with rights and obligations labels according to the component position information in the sentence labels; If it is determined that the component position information of the multi-type key sentence data is in a passive state, marking the multi-type key sentence data with a rights label; If it is determined that the component position information of the multi-type key sentence data is in an active state, the multi-type key sentence data is marked with mandatory labels; Judging the advantages and disadvantages of the clause texts of the multiple types of key sentence data according to the clause types and rights and obligations labels of the multiple types of key sentence data; If it is determined that the multi-type key sentence data is advantageous clause text data, obtaining advantageous clause weights; If it is determined that the multi-type key sentence data is inferior clause text data, obtaining inferior clause weights; The text risk data is calculated and acquired according to the weight of the advantageous clause and the weight of the disadvantageous clause, and the text risk data is input into a risk judgment module.
[0010] Furthermore, the step of obtaining historical cooperation data to determine the capability weight and credit weight of the target partner, and calculating and obtaining comprehensive risk data based on the capability weight and credit weight specifically includes: Obtain the capability data of both parties from the preset historical cooperation database and perform correlation processing based on the Bayesian probability algorithm to obtain the capability data of the target partner, the capability data of the party itself and the capability data of other reference partners related to the current contract; According to the preset capability grading standard, the target partner capability data, the own capability data and the other reference partner capability data are graded to obtain the target partner capability index, the own capability index and the other reference partner capability index respectively; Comparing the target partner's capability index with the capability indexes of other reference partners according to the capability categories to calculate the target partner's performance capability weight; Compare the capability index of the party with the capability index of the target partner and the capability index of other reference partners respectively, so as to calculate the negotiation capability weight of the target partner; Then, the target partner credit data is obtained from the preset historical cooperation database, and the impact judgment is made on the credit data of all the target partners respectively; If the target partner's credit data is determined to be positive data, a positive credit index is obtained; If the target partner's credit data is determined to be negative data, a negative credit index is obtained; Credit grading the credit index according to the preset credit grading standard to obtain the credit weight of the target partner; The comprehensive risk data is calculated and obtained according to the target partner's performance ability weight, the target partner's negotiation ability weight and the target partner's credit weight, and the comprehensive risk data is input into the risk judgment module.
[0011] Furthermore, the step of judging the text risk data and the comprehensive risk data according to a preset risk threshold to obtain a risk modification item specifically includes: Calculate the text risk probability based on the text risk data to obtain the text risk clause items; Performing risk grading on the text risk clause items according to a preset risk grading standard, so as to label the text risk clause items with text risk level labels; Determine whether the text risk level label is greater than a preset risk threshold; If it is determined that the text risk level label is greater than a preset risk threshold, the corresponding text risk clause item is input into the modification suggestion module as a risk modification item; If it is determined that the text risk level label is less than or equal to the preset risk threshold, the corresponding text risk clause item is input into the modification suggestion module as an advantage and disadvantage modification item; Calculate the comprehensive risk probability based on the comprehensive risk data to obtain comprehensive risk terms; Performing risk grading on the comprehensive risk clause items according to the preset risk grading standard, so as to label the comprehensive risk clause items with comprehensive risk level labels; Determine whether the comprehensive risk level label is greater than a preset risk threshold; If it is determined that the comprehensive risk level label is greater than the preset risk threshold, the corresponding comprehensive risk clause item is input into the modification suggestion module as a risk modification item; If it is determined that the comprehensive risk level label is less than or equal to the preset risk threshold, the corresponding comprehensive risk clause item is input into the modification suggestion module as an acceptance modification item.
[0012] Furthermore, the step of generating modification suggestions and final risk warning items according to the risk modification items specifically includes: Extracting features of the risk modification items according to a multimodal deep convolution algorithm to obtain target risk features, wherein the multimodal deep convolution algorithm includes a multimodal data conversion layer and a deep convolution layer; Inputting the target risk feature into an attention enhancement layer to obtain a target risk attention enhancement feature, wherein the attention enhancement layer includes a context feature enhancement layer and a dynamic feature enhancement layer; The context feature enhancement layer obtains context feature weights according to a context perception mechanism to perform context enhancement feature embedding on the target risk feature; The dynamic feature enhancement layer obtains the dynamic similarity weight according to the dynamic adjustment mechanism to dynamically enhance the feature embedding of the target risk feature; According to the target risk attention enhancement feature, a final risk prompt item and modification suggestion text data are obtained; Adjusting the modification suggestion text data according to the clause advantage maximization algorithm and making a feasibility judgment; If it is determined that the feasibility of the modified suggestion text data after adjustment is greater than or equal to a preset acceptable threshold, the modified suggestion text data is used as a first modified suggestion and is output as a prediction result together with the final risk warning item.
[0013] Furthermore, the step of adjusting the modification suggestion text data according to the clause advantage maximization algorithm and making a feasibility judgment also includes: Adjust the modification suggestion text data according to the advantage maximization algorithm and make feasibility judgment; If it is determined that the feasibility of the modified suggestion text data after adjustment is less than a preset acceptable threshold, gradient adjustment is performed on the modified suggestion text data according to a clause acceptance gradient algorithm; When the feasibility of the modification suggestion text data is greater than or equal to a preset acceptable threshold, the modification suggestion text data is used as a second modification suggestion and is output as a prediction result together with the final risk warning item.
[0014] The present invention proposes a contract comprehensive risk prediction system, comprising: A text risk acquisition module is used to acquire contract text data and perform word segmentation processing, classify and weight the contract text data after word segmentation processing to obtain clause weights, and calculate and acquire text risk data based on the clause weights; A comprehensive risk acquisition module is used to acquire historical cooperation data to determine the capability weight and credit weight of the target partner, and calculate and acquire comprehensive risk data based on the capability weight and credit weight; The modification suggestion module is used to judge the text risk data and the comprehensive risk data according to a preset risk threshold to obtain risk modification items, and to generate modification suggestions and final risk prompt items according to the risk modification items.
[0015] The present invention also provides a storage medium, which stores one or more programs, and when the programs are executed by a processor, the above-mentioned comprehensive contract risk prediction method is implemented.
[0016] The present invention also provides a computer device, the computer device comprising a memory and a processor, wherein: The memory is used to store computer programs; When the processor is used to execute the computer program stored in the memory, the above-mentioned comprehensive contract risk prediction method is implemented. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] Figure 1 A flowchart of a method for predicting comprehensive contract risks proposed in the first embodiment of the present invention; Figure 2 A flow chart of a method for predicting comprehensive contract risks proposed in a second embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of the comprehensive contract risk prediction system proposed in the third embodiment of the present invention.
[0018] The following specific implementation manner will further illustrate the present invention in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION
[0019] In order to facilitate understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Several embodiments of the present invention are provided in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.
[0020] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.
[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.
[0022] See also Figure 1 , which is a flow chart of a contract comprehensive risk prediction method proposed in the first embodiment of the present invention, and the contract comprehensive risk prediction method includes steps S01 to S03, wherein: Step S01: Acquire contract text data and perform word segmentation processing, classify and weight the contract text data after word segmentation processing to obtain clause weights, and obtain text risk data based on the clause weights; It should be noted that in this embodiment, the word segmentation process is used to divide the single word data in the contract text data and mark the components of the single word data in the sentence, specifically to obtain the contract text data and perform preprocessing, the preprocessing includes text cleaning, text format normalization and text detection filtering; The text cleaning includes removing special characters, blank characters and stop words in the contract text data; The text format normalization includes respectively formatting the Chinese text, Western text, digital text and time text in the contract text data according to the preset contract text specification; The preset contract text specifications in this embodiment include text font format, Western uppercase and lowercase format, digital counting format and time sequence format; The text detection and filtering includes detecting and filtering the template application paragraphs, standard statement paragraphs and regulatory reference paragraphs in the contract text data; Then, the pre-processed contract text data is subjected to text segmentation to obtain single sentence data, and the single sentence data is subjected to word segmentation processing, wherein the word segmentation processing includes word segmentation and sentence component labeling; The word segmentation includes identifying and determining whether a multi-character combination in the single sentence data is a single word; If it is determined that the multi-character combination is not a single word, the multi-character combination is divided into secondary characters according to the secondary segmentation method and a single word determination is performed; If it is determined that the multiple characters are combined into a single word, extracting the single word to combine into single word data; The sentence component tagging includes dividing the single sentence data into sentence components, obtaining sentence component character segments in the single sentence data, and matching the single word data with the sentence component character segments to tag the single word data with a sentence label, wherein the sentence label includes component position information of the single word data in the corresponding single sentence data and paragraph position information of the single sentence data in the contract text data; Input the single word data after word segmentation into the classification weighting module; The classification weighting process described in this embodiment is used to classify the clauses in the contract text data and add weights to them respectively, specifically, keyword matching is performed on single word data according to a preset keyword database to obtain key word data, and key sentence data is located and obtained according to the paragraph position information in the sentence tag; The preset keyword database in this embodiment includes contract subject keywords, contract item keywords, contract element keywords, contract rights and obligations keywords, and contract value keywords; Classifying the key sentence data into clause types according to the keyword types to obtain multiple types of key sentence data; Then, all the multi-type key sentence data are labeled with rights and obligations labels according to the component position information in the sentence labels; If it is determined that the component position information of the multi-type key sentence data is in a passive state, marking the multi-type key sentence data with a rights label; If it is determined that the component position information of the multi-type key sentence data is in an active state, the multi-type key sentence data is marked with mandatory labels; Judging the advantages and disadvantages of the clause texts of the multiple types of key sentence data according to the clause types and rights and obligations labels of the multiple types of key sentence data; If it is determined that the multi-type key sentence data is advantageous clause text data, obtaining advantageous clause weights; If it is determined that the multi-type key sentence data is inferior clause text data, obtaining inferior clause weights; The text risk data is calculated and obtained according to the weight of the advantageous clauses and the weight of the disadvantageous clauses, and the text risk data is input into the risk judgment module. In this embodiment, the text risk data is the text data of all risky clauses in the contract text data and the risk weight of the text data.
[0023] Step S02: Acquire historical cooperation data to determine the capability weight and credit weight of the target partner, and calculate and acquire comprehensive risk data based on the capability weight and the credit weight; It should be noted that the historical cooperation data in this embodiment refers to the partner data of the current contract, the data of the party itself, and the data of other partners of the same type as the partner of the current contract. Specifically, the capability data of both parties of the cooperation are obtained from a preset historical cooperation database and correlation processing is performed according to the Bayesian probability algorithm to obtain the capability data of the target partner, the capability data of the party itself, and the capability data of other reference partners related to the current contract; According to the preset capability grading standard, the target partner capability data, the own capability data and the other reference partner capability data are graded to obtain the target partner capability index, the own capability index and the other reference partner capability index respectively; The capability grading standard in this embodiment includes strong capability level, medium capability level and weak capability level. The strong capability level is that the capability related to the current contract project is in the top 30% of the overall industry ranking, the medium capability level is that the capability related to the current contract project is in the top 30% to 70% of the overall industry ranking, and the weak capability level is that the capability related to the current contract project is in the bottom 30% of the overall industry ranking; Comparing the target partner's capability index with the capability indexes of other reference partners according to the capability categories to calculate the target partner's performance capability weight; Compare the capability index of the party with the capability index of the target partner and the capability index of other reference partners respectively, so as to calculate the negotiation capability weight of the target partner; Then, the target partner credit data is obtained from the preset historical cooperation database, and the impact judgment is made on the credit data of all the target partners respectively; If the target partner's credit data is determined to be positive data, a positive credit index is obtained; If the target partner's credit data is determined to be negative data, a negative credit index is obtained; Credit grading the credit index according to the preset credit grading standard to obtain the credit weight of the target partner; The comprehensive risk data is calculated and obtained based on the target partner's performance ability weight, the target partner's negotiation ability weight and the target partner's credit weight, and the comprehensive risk data is input into the risk judgment module. In this embodiment, the comprehensive risk data is all the ability or credit data with risk problems in the partner data of the current contract and the risk weight of the ability or credit data.
[0024] Step S03: judging the text risk data and the comprehensive risk data according to a preset risk threshold to obtain risk modification items, and generating modification suggestions and final risk warning items according to the risk modification items; It should be noted that in this embodiment, the text risk probability calculation is performed based on the text risk data to obtain the text risk clause items, and the risk modification items are all the contract clause items that have risks and need to be modified; Performing risk grading on the text risk clause items according to a preset risk grading standard, so as to label the text risk clause items with text risk level labels; The preset risk classification standards in this embodiment include high risk level, medium risk level and low risk level, wherein the high risk level means that there are inferior rights clauses and inferior obligations clauses and the number of superior rights clauses is less than the superior obligations clauses, the medium risk level means that there are inferior rights clauses or inferior obligations clauses and the number of superior rights clauses is less than the superior obligations clauses, and the low risk level means that there are inferior rights clauses or inferior obligations clauses or the number of superior rights clauses is less than the superior obligations clauses; Determine whether the text risk level label is greater than a preset risk threshold; The preset risk threshold in this embodiment is the medium risk level label value; If it is determined that the text risk level label is greater than a preset risk threshold, the corresponding text risk clause item is input into the modification suggestion module as a risk modification item; If it is determined that the text risk level label is less than or equal to the preset risk threshold, the corresponding text risk clause item is input into the modification suggestion module as an advantage and disadvantage modification item; Calculate the comprehensive risk probability based on the comprehensive risk data to obtain comprehensive risk terms; Performing risk grading on the comprehensive risk clause items according to the preset risk grading standard, so as to label the comprehensive risk clause items with comprehensive risk level labels; Determine whether the comprehensive risk level label is greater than a preset risk threshold; If it is determined that the comprehensive risk level label is greater than the preset risk threshold, the corresponding comprehensive risk clause item is input into the modification suggestion module as a risk modification item; If it is determined that the comprehensive risk level label is less than or equal to the preset risk threshold, the corresponding comprehensive risk clause item is input into the modification suggestion module as an acceptance modification item; Extracting features of the risk modification items according to a multimodal deep convolution algorithm to obtain target risk features, wherein the multimodal deep convolution algorithm includes a multimodal data conversion layer and a deep convolution layer; Inputting the target risk feature into an attention enhancement layer to obtain a target risk attention enhancement feature, wherein the attention enhancement layer includes a context feature enhancement layer and a dynamic feature enhancement layer; The context feature enhancement layer obtains context feature weights according to a context perception mechanism to perform context enhancement feature embedding on the target risk feature; The dynamic feature enhancement layer obtains the dynamic similarity weight according to the dynamic adjustment mechanism to dynamically enhance the feature embedding of the target risk feature; According to the target risk attention enhancement feature, a final risk prompt item and modification suggestion text data are obtained; Adjusting the modification suggestion text data according to the clause advantage maximization algorithm and making a feasibility judgment; If it is determined that the feasibility of the modified suggestion text data after adjustment is greater than or equal to a preset acceptable threshold, the modified suggestion text data is used as a first modified suggestion and outputted as a prediction result together with the final risk warning item; The preset acceptable threshold in this embodiment is the target partner's negotiation capability weight value; If it is determined that the feasibility of the modified suggestion text data after adjustment is less than a preset acceptable threshold, gradient adjustment is performed on the modified suggestion text data according to a clause acceptance gradient algorithm; When the feasibility of the modification suggestion text data is greater than or equal to a preset acceptable threshold, the modification suggestion text data is used as a second modification suggestion and is output as a prediction result together with the final risk warning item.
[0025] In summary, according to the above-mentioned comprehensive contract risk prediction method, by first obtaining the clause weights in the contract text data, calculating the text risk data, and then judging the target partner's ability weight and credit weight based on the historical cooperation data, and calculating the comprehensive risk data, the risk modification items are obtained based on the text risk data and the comprehensive risk data. Not only is the risk judgment of the text trap made at the text level, but also the multi-dimensional consideration of the partner's performance ability and overall credit is made to judge the risk clauses of the contract, while taking into account the strengths and weaknesses of the contract partners and the acceptance of the negotiation. The present invention greatly improves the accuracy of contract risk prediction and the feasibility of modification suggestions. The contract text data is obtained and segmented, and the contract text data after the segmentation is classified and weighted to obtain the clause weight, and the text risk data is obtained according to the clause weight, so as to avoid the cooperation risk that may be caused by text traps or disadvantageous clauses, and the historical cooperation data is obtained to judge the ability weight and credit weight of the target partner, and the comprehensive risk data is obtained according to the ability weight and the credit weight. The multi-dimensional considerations are made from the partners' performance ability and overall credit, so as to avoid the contract breach caused by the legal risk and financial risk of the partner, and the text risk data and the comprehensive risk data are judged according to the preset risk threshold to obtain the risk modification items, so as to generate the modification suggestions and the final risk warning items according to the risk modification items, and at the same time of multi-dimensional risk warnings, the clause modification suggestions are made according to the advantages and disadvantages of the ability and the negotiation position, and at the same time, the acceptance of the cooperation parties is further guaranteed. The present invention greatly improves the accuracy of contract risk prediction and the feasibility of modification suggestions.
[0026] See also Figure 2 , which is a flow chart of a contract comprehensive risk prediction method proposed in the second embodiment of the present invention, and the contract comprehensive risk prediction method includes steps S11 to S16, wherein: Step S11: obtaining contract text data and preprocessing it, then performing text segmentation on the preprocessed contract text data to obtain single sentence data, performing word segmentation on the single sentence data and then inputting it into a classification weighting module; Step S12: keyword matching is performed on single word data according to a preset keyword database to obtain key word data, and key sentence data is located and obtained according to the paragraph position information in the sentence tag, and the key sentence data is classified into clause types according to the keyword type to obtain multiple types of key sentence data, and then rights and obligations labels are marked and judged for all multiple types of key sentence data according to the component position information in the sentence tag, and the advantages and disadvantages of the clause text of the multiple types of key sentence data are judged according to the clause type and rights and obligations labels of the multiple types of key sentence data to obtain clause weights, and the text risk data is calculated and obtained; Step S13: Acquire the capability data of both parties from the preset historical cooperation database and perform correlation processing according to the Bayesian probability algorithm, perform capability grading according to the preset capability grading standard, perform capability comparison calculation to obtain the performance capability weight and negotiation capability weight of the target partner, and then obtain the credit data of the target partner from the preset historical cooperation database to make an impact judgment, perform credit grading on the credit index according to the preset credit grading standard to obtain the credit weight of the target partner, so as to calculate and obtain comprehensive risk data; Step S14: Calculate the risk probability based on the text risk data and the comprehensive risk data to obtain risk clause items, perform risk grading based on the preset risk grading standard, label the risk clause items with risk level tags, and determine whether the risk level tag is greater than the preset risk threshold. If the risk level tag is greater than the preset risk threshold, the corresponding risk clause item is used as a risk modification item; Step S15: extracting features of the risk modification items according to the multimodal deep convolution algorithm to obtain target risk features, inputting the target risk features into the attention enhancement layer to obtain target risk attention enhancement features, obtaining final risk prompt items and modification suggestion text data according to the target risk attention enhancement features, adjusting the modification suggestion text data according to the clause advantage maximization algorithm, and making a feasibility judgment; Step S16: If the feasibility of the modified suggestion text data after adjustment is determined to be greater than or equal to the preset acceptable threshold, the modified suggestion text data is used as the first modified suggestion and is output as a prediction result together with the final risk warning item. If the feasibility of the modified suggestion text data after adjustment is determined to be less than the preset acceptable threshold, the modified suggestion text data is gradient-adjusted according to the clause acceptance gradient algorithm. When the feasibility of the modified suggestion text data is greater than or equal to the preset acceptable threshold, the modified suggestion text data is used as the second modified suggestion and is output as a prediction result together with the final risk warning item.
[0027] In summary, according to the above-mentioned comprehensive contract risk prediction method, by first obtaining the clause weights in the contract text data, calculating the text risk data, and then judging the target partner's ability weight and credit weight based on the historical cooperation data, and calculating the comprehensive risk data, the risk modification items are obtained based on the text risk data and the comprehensive risk data. Not only is the risk judgment of the text trap made at the text level, but also the multi-dimensional consideration of the partner's performance ability and overall credit is made to judge the risk clauses of the contract, while taking into account the strengths and weaknesses of the contract partners and the acceptance of the negotiation. The present invention greatly improves the accuracy of contract risk prediction and the feasibility of modification suggestions. The contract text data is obtained and segmented, and the contract text data after the segmentation is classified and weighted to obtain the clause weight, and the text risk data is obtained according to the clause weight, so as to avoid the cooperation risk that may be caused by text traps or disadvantageous clauses, and the historical cooperation data is obtained to judge the ability weight and credit weight of the target partner, and the comprehensive risk data is obtained according to the ability weight and the credit weight. The multi-dimensional considerations are made from the partners' performance ability and overall credit, so as to avoid the contract breach caused by the legal risk and financial risk of the partner, and the text risk data and the comprehensive risk data are judged according to the preset risk threshold to obtain the risk modification items, so as to generate the modification suggestions and the final risk warning items according to the risk modification items, and at the same time of multi-dimensional risk warnings, the clause modification suggestions are made according to the advantages and disadvantages of the ability and the negotiation position, and at the same time, the acceptance of the cooperation parties is further guaranteed. The present invention greatly improves the accuracy of contract risk prediction and the feasibility of modification suggestions.
[0028] See also Figure 3 , which is a schematic diagram of the structure of a contract comprehensive risk prediction system proposed in the third embodiment of the present invention, and the system includes: The text risk acquisition module 10 is used to acquire contract text data and perform word segmentation processing, classify and weight the contract text data after word segmentation processing to obtain clause weights, and calculate and acquire text risk data based on the clause weights; The comprehensive risk acquisition module 20 is used to acquire historical cooperation data to determine the capability weight and credit weight of the target partner, and calculate and acquire comprehensive risk data based on the capability weight and credit weight; The modification suggestion module 30 is used to judge the text risk data and the comprehensive risk data according to a preset risk threshold to obtain risk modification items, and to generate modification suggestions and final risk prompt items according to the risk modification items.
[0029] The present invention also proposes a computer storage medium on which one or more programs are stored, and when the program is executed by a processor, the above-mentioned contract comprehensive risk prediction method is implemented.
[0030] The present invention also proposes a computer device, comprising a memory and a processor, wherein the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory to implement the above-mentioned contract comprehensive risk prediction method.
[0031] Those skilled in the art will appreciate that the logic and / or steps represented in the flowchart or otherwise described herein, for example, may be considered as an ordered list of executable instructions for implementing logical functions, and may be specifically implemented in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For purposes of this specification, "computer-readable medium" may be any device that can contain storage, communication, propagation or transmission of a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.
[0032] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.
[0033] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or a combination thereof: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.
[0034] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.
[0035] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.
Claims
1. A method for predicting comprehensive contract risks, characterized in that: include: Acquire contract text data and perform word segmentation processing, perform classification and weighted processing on the contract text data after word segmentation processing to obtain clause weights, and obtain text risk data based on the clause weights; Obtain historical cooperation data to determine the capability weight and credit weight of the target partner, and calculate and obtain comprehensive risk data based on the capability weight and credit weight; The text risk data and the comprehensive risk data are judged according to a preset risk threshold to obtain risk modification items, so as to generate modification suggestions and final risk prompt items according to the risk modification items.
2. The method for predicting comprehensive contract risks according to claim 1, characterized in that: The step of obtaining the contract text data and performing word segmentation processing specifically includes: Acquire contract text data and perform preprocessing, wherein the preprocessing includes text cleaning, text format normalization, and text detection and filtering; The text cleaning includes removing special characters, blank characters and stop words in the contract text data; The text format normalization includes respectively formatting the Chinese text, Western text, digital text and time text in the contract text data according to the preset contract text specification; The text detection and filtering includes detecting and filtering the template application paragraphs, standard statement paragraphs and regulatory reference paragraphs in the contract text data; Then, the pre-processed contract text data is subjected to text segmentation to obtain single sentence data, and the single sentence data is subjected to word segmentation processing, wherein the word segmentation processing includes word segmentation and sentence component labeling; The word segmentation includes identifying and determining whether a multi-character combination in the single sentence data is a single word; If it is determined that the multi-character combination is not a single word, the multi-character combination is divided into secondary characters according to the secondary segmentation method and a single word determination is performed; If it is determined that the multiple characters are combined into a single word, extracting the single word to combine into single word data; The sentence component tagging includes dividing the single sentence data into sentence components, obtaining sentence component character segments in the single sentence data, and matching the single word data with the sentence component character segments to tag the single word data with a sentence label, wherein the sentence label includes component position information of the single word data in the corresponding single sentence data and paragraph position information of the single sentence data in the contract text data; The single word data after word segmentation processing is input into the classification weighting module.
3. The method for predicting comprehensive contract risks according to claim 2, characterized in that: After the step of inputting the single word data after word segmentation processing into the classification weighting module, the method further includes: Perform keyword matching on single word data according to a preset keyword database to obtain key word data, and locate and obtain key sentence data according to paragraph position information in sentence tags; Classifying the key sentence data into clause types according to the keyword types to obtain multiple types of key sentence data; Then, all the multi-type key sentence data are labeled with rights and obligations labels according to the component position information in the sentence labels; If it is determined that the component position information of the multi-type key sentence data is in a passive state, marking the multi-type key sentence data with a rights label; If it is determined that the component position information of the multi-type key sentence data is in an active state, the multi-type key sentence data is marked with mandatory labels; Judging the advantages and disadvantages of the clause texts of the multiple types of key sentence data according to the clause types and rights and obligations labels of the multiple types of key sentence data; If it is determined that the multi-type key sentence data is advantageous clause text data, obtaining advantageous clause weights; If it is determined that the multi-type key sentence data is inferior clause text data, obtaining inferior clause weights; The text risk data is calculated and acquired according to the weight of the advantageous clause and the weight of the disadvantageous clause, and the text risk data is input into a risk judgment module.
4. The method for predicting comprehensive contract risks according to claim 1, characterized in that: The step of obtaining historical cooperation data to determine the capability weight and credit weight of the target partner, and calculating and obtaining comprehensive risk data based on the capability weight and credit weight, specifically includes: Obtain the capability data of both parties from the preset historical cooperation database and perform correlation processing based on the Bayesian probability algorithm to obtain the capability data of the target partner, the capability data of the party itself and the capability data of other reference partners related to the current contract; According to the preset capability grading standard, the target partner capability data, the own capability data and the other reference partner capability data are graded to obtain the target partner capability index, the own capability index and the other reference partner capability index respectively; Comparing the target partner's capability index with the capability indexes of other reference partners according to the capability categories to calculate the target partner's performance capability weight; Compare the capability index of the party with the capability index of the target partner and the capability index of other reference partners respectively, so as to calculate the negotiation capability weight of the target partner; Then, the target partner credit data is obtained from the preset historical cooperation database, and the impact judgment is made on the credit data of all the target partners respectively; If the target partner's credit data is determined to be positive data, a positive credit index is obtained; If the target partner's credit data is determined to be negative data, a negative credit index is obtained; Credit grading the credit index according to the preset credit grading standard to obtain the credit weight of the target partner; The comprehensive risk data is calculated and obtained according to the target partner's performance ability weight, the target partner's negotiation ability weight and the target partner's credit weight, and the comprehensive risk data is input into the risk judgment module.
5. The method for predicting comprehensive contract risks according to claim 1, characterized in that: The step of judging the text risk data and the comprehensive risk data according to a preset risk threshold to obtain a risk modification item specifically includes: Calculate the text risk probability based on the text risk data to obtain the text risk clause items; Performing risk grading on the text risk clause items according to a preset risk grading standard, so as to label the text risk clause items with text risk level labels; Determine whether the text risk level label is greater than a preset risk threshold; If it is determined that the text risk level label is greater than a preset risk threshold, the corresponding text risk clause item is input into the modification suggestion module as a risk modification item; If it is determined that the text risk level label is less than or equal to the preset risk threshold, the corresponding text risk clause item is input into the modification suggestion module as an advantage and disadvantage modification item; Calculate the comprehensive risk probability based on the comprehensive risk data to obtain comprehensive risk terms; Performing risk grading on the comprehensive risk clause items according to the preset risk grading standard, so as to label the comprehensive risk clause items with comprehensive risk level labels; Determine whether the comprehensive risk level label is greater than a preset risk threshold; If it is determined that the comprehensive risk level label is greater than the preset risk threshold, the corresponding comprehensive risk clause item is input into the modification suggestion module as a risk modification item; If it is determined that the comprehensive risk level label is less than or equal to the preset risk threshold, the corresponding comprehensive risk clause item is input into the modification suggestion module as an acceptance modification item.
6. The method for predicting comprehensive contract risks according to claim 1, characterized in that: The step of generating modification suggestions and final risk warning items according to the risk modification items specifically includes: Extracting features of the risk modification items according to a multimodal deep convolution algorithm to obtain target risk features, wherein the multimodal deep convolution algorithm includes a multimodal data conversion layer and a deep convolution layer; Inputting the target risk feature into an attention enhancement layer to obtain a target risk attention enhancement feature, wherein the attention enhancement layer includes a context feature enhancement layer and a dynamic feature enhancement layer; The context feature enhancement layer obtains context feature weights according to a context perception mechanism to perform context enhancement feature embedding on the target risk feature; The dynamic feature enhancement layer obtains the dynamic similarity weight according to the dynamic adjustment mechanism to dynamically enhance the feature embedding of the target risk feature; According to the target risk attention enhancement feature, a final risk prompt item and modification suggestion text data are obtained; Adjusting the modification suggestion text data according to the clause advantage maximization algorithm and making a feasibility judgment; If it is determined that the feasibility of the modified suggestion text data after adjustment is greater than or equal to a preset acceptable threshold, the modified suggestion text data is used as a first modified suggestion and is output as a prediction result together with the final risk warning item.
7. The method for predicting comprehensive contract risks according to claim 6, characterized in that: The step of adjusting the modification suggestion text data according to the clause advantage maximization algorithm and making a feasibility judgment also includes: Adjust the modification suggestion text data according to the advantage maximization algorithm and make feasibility judgment; If it is determined that the feasibility of the modified suggestion text data after adjustment is less than a preset acceptable threshold, gradient adjustment is performed on the modified suggestion text data according to a clause acceptance gradient algorithm; When the feasibility of the modification suggestion text data is greater than or equal to a preset acceptable threshold, the modification suggestion text data is used as a second modification suggestion and is output as a prediction result together with the final risk warning item.
8. A contract comprehensive risk prediction system, characterized in that: include: A text risk acquisition module is used to acquire contract text data and perform word segmentation processing, classify and weight the contract text data after word segmentation processing to obtain clause weights, and calculate and acquire text risk data based on the clause weights; A comprehensive risk acquisition module is used to acquire historical cooperation data to determine the capability weight and credit weight of the target partner, and calculate and acquire comprehensive risk data based on the capability weight and credit weight; The modification suggestion module is used to judge the text risk data and the comprehensive risk data according to a preset risk threshold to obtain risk modification items, and to generate modification suggestions and final risk prompt items according to the risk modification items.
9. A storage medium, characterized in that: The storage medium stores one or more programs, which, when executed by a processor, implement the contract comprehensive risk prediction method as described in any one of claims 1-7.
10. A computer device, characterized in that: The computer device comprises a memory and a processor, wherein: The memory is used to store computer programs; When the processor is used to execute the computer program stored in the memory, it implements the contract comprehensive risk prediction method described in any one of claims 1-7.
Citation Information
Cited By
Contract risk prediction method and device, equipment and storage medium
CN120875589A
A contract risk prediction method, device and equipment and storage medium
CN120875589B
