Intelligent contract full-automatic circulation and effectiveness method

Abstract

The invention relates to the technical field of artificial intelligence, in particular to a full-automatic circulation and effectiveness method for an intelligent contract. Comprising the steps of contract information input, contract entry verification and anomaly detection, automatic generation of a contract draft, check and adjustment of contract information, automatic examination and approval of a contract, anomaly processing and introduction of an automatic feedback mechanism, automatic signature of the contract, automatic processing of contract effectiveness and dynamic contract management. Manual intervention is reduced in the process from information entry to approval and effectiveness of the contract, the working efficiency is greatly improved, and manual errors are reduced; according to the method, an SVM model is combined with a classification algorithm, forward greedy search and weighting processing are used, important features are dynamically selected and weighted according to different contract types, the classification effect is enhanced, the importance of each feature on contract term classification is quantified by introducing an evaluation function based on information gain, and the classification efficiency is improved. And the clause classification accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and specifically to a fully automated circulation and effectiveness method for smart contracts. Background Art

[0002] With the rapid development of digitalization and information technology, contract management has gradually transitioned from the traditional paper contract model to the electronic contract and smart contract model. The traditional contract circulation and effectiveness process relies on manual operations. The contract drafting and effectiveness cycle is relatively long, especially when multiple parties sign. Communication and confirmation will further delay the time. In addition, when manually entering and proofreading contract information, it is easy to have problems such as mis-input, missing items, and irregular clause formats. General contract management has problems such as high recognition error rate and poor flexibility. Special clauses in the contract may be omitted or misclassified, resulting in key clauses in the draft contract cannot be correctly identified or processed, and high-risk clauses such as compensation and breach of contract cannot be accurately identified, which increases the legal risks and commercial disputes in contract execution. In addition, it is impossible to flexibly respond to different types of contract clause classifications, resulting in insufficient versatility and scalability of the system, and it cannot adapt to the diversified needs of contracts in various industries. Summary of the invention

[0003] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a fully automated circulation and effectiveness method for intelligent contracts. In order to solve the problems of high error rate and low efficiency in the traditional contract circulation and effectiveness process, the present invention reduces manual intervention in the process from information entry to approval and effectiveness through a fully automated process, greatly improves work efficiency, quickly completes the steps of contract draft generation, verification, approval, and signing, shortens the circulation cycle of the contract, and reduces manual errors; in view of the problems of high recognition error rate and poor flexibility in general contract management, the present invention uses an SVM model in combination with a classification algorithm, automatically and accurately identifies and classifies special terms by setting classification labels, customizes or expands the classification rules and labels of special terms according to different types of contracts, and responds to different types of contracts and industry needs, and uses forward greedy search and weighted processing to dynamically select and weight important features according to different contract types, thereby enhancing the classification effect, and introduces an evaluation function based on information gain to quantify the importance of each feature for the classification of contract terms, thereby improving the accuracy of term classification.

[0004] The technical solution adopted by the present invention is as follows: The present invention provides a fully automated circulation and effectiveness method of a smart contract, the method comprising the following steps:

[0005] Step S1: Contract information entry;

[0006] Step S2: Verify contract items and perform anomaly detection;

[0007] Step S3: Automatically generate a draft contract;

[0008] Step S4: Check and adjust contract information;

[0009] Step S5: Automatically approve the contract;

[0010] Step S6: Perform exception handling and introduce automatic feedback mechanism;

[0011] Step S7: Automatically sign the contract;

[0012] Step S8: Automatic processing of contract effectiveness;

[0013] Step S9: Dynamic contract management.

[0014] Furthermore, in step S1, a contract draft is defined, and the winning bid information is obtained from the bidding results and the winning bid notice, the winning bid information including the winning bid items and the contract parties, and the winning bid information is preprocessed and format converted using a data preprocessing algorithm to obtain the winning bid data;

[0015] Furthermore, in step S2, a set of contract templates is predefined, including contract items and corresponding formats, and a data verification algorithm is used to check whether the winning items are consistent with the contract items through an item comparison mechanism. If there is any inconsistency, it is marked and a difference report is generated;

[0016] Furthermore, in step S3, a template generation algorithm is used to fill in the contract template according to the winning bid data to obtain a contract draft. For special contracts, special terms are identified through a classification algorithm, and the contract template is adjusted before filling in;

[0017] Furthermore, in step S4, the draft contract is checked using a text comparison algorithm, and when inconsistencies are found, a comparison report is generated and the draft contract is modified based on the comparison report;

[0018] Furthermore, in step S6, an anomaly detection algorithm is used in the contract approval process to identify potential errors and defects in the draft contract, and automatic correction suggestions are provided through rule matching. An automatic feedback mechanism is introduced to feed back the automatic correction suggestions to the draft contract for correction to obtain a formal contract;

[0019] Furthermore, in step S7, after the contract is approved, the digital signature process is executed, and the contract document is digitally signed through a digital signature and encryption algorithm to obtain a signed contract;

[0020] Furthermore, in step S8, the automatic effectiveness mechanism is referenced, and the contract is marked as effective after signing and archived in the contract library.

[0021] Furthermore, in step S3, special terms are identified by a classification algorithm, which specifically includes the following steps:

[0022] Step S31: Setting classification targets, classifying contract terms, and setting classification labels for contract terms, including compensation terms, breach of contract terms, and term terms;

[0023] Step S32: preparing training data, preparing three training data sets for three classification labels from existing contract terms, each data set containing sample data;

[0024] Step S33: introducing the SVM model, extracting text features from the draft contract, and converting it into a vector form;

[0025] Step S34: weighted processing, assigning weights according to the classification labels of the contract terms, and introducing weighted nearest neighbor feature selection to perform weighted processing on the sample data;

[0026] Step S35: Model training, screening out important features in the contract terms through forward greedy search, and starting to train the SVM model;

[0027] Step S36: Model optimization, using the alternating direction multiplier method to optimize the SVM model parameters;

[0028] Step S37: Identification and classification: input the pre-processed draft contract terms into the trained SR-SVM model, classify them according to the characteristics of the terms, and identify the classification labels of the contract terms;

[0029] Further, step S34 specifically includes the following contents:

[0030] Calculate neighborhood distance: For each sample data, select k nearest neighbor samples and calculate the neighborhood distance respectively;

[0031] Calculate standard deviation: Calculate the standard deviation of the class label for each sample as a measure of sample quality;

[0032] Assign weights: Weight the samples according to the standard deviation of the class labels of each sample. The samples with high class label standard deviation have lower weights, and the samples with low class label standard deviation have higher weights.

[0033] Furthermore, in step S35, important features in the contract terms are screened out through forward greedy search, which specifically includes the following steps:

[0034] Step S351: define the initial state, define the initial state of the root node as the set of all contract terms, each contract term corresponds to a feature, each root node represents a feature set, and a subset of the feature set is a feature combination;

[0035] Step S352: introducing an evaluation function, and using an evaluation function based on information gain to calculate the importance of each feature in the classification of contract terms;

[0036] Step S353: Select feature combinations, use the evaluation function to evaluate all feature combinations in the root node, select m feature combinations with the highest evaluation function values ​​as the child nodes of the second layer, and for each child node of the second layer, use the evaluation function again to select m optimal child nodes to generate m×m child nodes of the third layer;

[0037] Step S354: iterative operation, setting the depth d of the tree, iterating according to step S353, generating m×m child nodes based on the optimal child node in each iteration, until the depth d of the tree is reached and the iteration is terminated, generating a feature selection tree, each leaf node of the feature selection tree contains a feature combination;

[0038] Step S355: Introduce a greedy algorithm, and use the greedy algorithm to generate the current optimal feature combination for each leaf node of the feature selection tree. Starting from the feature combination of the leaf node, combined with the feature set of the upper node, the current optimal feature combination is selected layer by layer;

[0039] Step S356: Select the optimal feature set, repeat the iteration step S355, record the current best feature combination obtained in each iteration, select the best feature combination layer by layer, and obtain the optimal feature set;

[0040] Step S357: Output the result, outputting the optimal feature set as the important features in the contract terms, including the combination of different types of contract terms;

[0041] Furthermore, step S352 specifically includes the following contents:

[0042] The evaluation function based on information gain selects the feature combination with the highest information gain. The formula of the evaluation function is as follows: ;

[0043] in, is the entropy of the set of all contract terms, Indicates the classification of contract terms The probability of ;

[0044] in, is the set of feature A in all contract terms The information gain on Let the value of feature A in S be When the subset is the total number of features, for The entropy of .

[0045] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0046] (1) Through the fully automated process, the process from contract information entry to approval and effectiveness has reduced manual intervention, greatly improved work efficiency, and quickly completed the steps of contract draft generation, verification, approval, and signing, shortening the contract circulation cycle and reducing manual errors.

[0047] (2) By using the SVM model in combination with the classification algorithm, special terms can be automatically and accurately identified and classified by setting classification labels. The classification rules and labels of special terms can be customized or extended according to different types of contracts to meet the needs of different contract types and industries. By using forward greedy search and weighted processing, important features can be dynamically selected and weighted according to different contract types, thereby enhancing the classification effect. By introducing an evaluation function based on information gain, the importance of each feature for the classification of contract terms can be quantified, thereby improving the accuracy of term classification. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A schematic diagram of a process for a fully automated circulation and validation method of a smart contract provided by the present invention;

[0049] Figure 2 A schematic diagram of screening out important features in contract terms through forward greedy search provided by the present invention.

[0050] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0052] In the description of the present invention, it should be understood that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as limiting the present invention.

[0053] Example 1, see Figure 1 ,The present invention provides a fully automated circulation and effectiveness method for smart contracts, the method comprising the following steps: Step S1: contract information input;

[0054] Step S2: Verify contract items and perform anomaly detection;

[0055] Step S3: Automatically generate a draft contract;

[0056] Step S4: Check and adjust contract information;

[0057] Step S5: Automatically approve the contract;

[0058] Step S6: Perform exception handling and introduce automatic feedback mechanism;

[0059] Step S7: Automatically sign the contract;

[0060] Step S8: Automatic processing of contract effectiveness;

[0061] Step S9: Dynamic contract management.

[0062] Example 2, see Figure 1 , this embodiment is based on the above embodiment. In step S1, a contract draft is defined, and the winning bid information is obtained from the bidding results and the winning bid notice. The winning bid information includes the winning bid items and the contract parties. The winning bid information is preprocessed and formatted using a data preprocessing algorithm to obtain the winning bid data;

[0063] In this embodiment, a contract draft is created, including general terms, legal requirements and standard formats, the latest bidding results and winning notices are obtained from the electronic bidding platform, and the winning information is parsed using a natural language processing algorithm to obtain the winning items, contract party information, contract amount, time frame and payment terms.

[0064] Embodiment 3: This embodiment is based on the above embodiment. In step S2, a set of contract templates is predefined, including contract items and corresponding formats. A data verification algorithm is used to check whether the winning items are consistent with the contract items through an item comparison mechanism. If there is any inconsistency, it is marked and a difference report is generated.

[0065] In this embodiment, a string matching algorithm is used to perform content consistency comparison of contract items, a regular expression-based format check is used, a hash table is used to store contract items in the template, hash comparison is performed through the contract item name and content, inconsistent items are quickly found, a difference report is generated in JSON format, and exported as an Excel file for manual review.

[0066] Embodiment 4: This embodiment is based on the above embodiment. In step S3, a template generation algorithm is used to fill in the contract template according to the winning bid data to obtain a contract draft. For special contracts, special terms are identified through a classification algorithm, and the contract template is adjusted before filling in.

[0067] Embodiment 5, based on the above embodiment, in step S4, the text comparison algorithm is used to check the draft contract, and when there is inconsistency, a comparison report is generated and the draft contract is modified according to the comparison report;

[0068] In this embodiment, the Jaccard similarity algorithm is used, the target similarity is preset, the contract draft is compared with the predefined contract template, the similarity score is calculated, the similarity score is marked for each inconsistent item, and after the contract draft is revised, a comparison is performed again, and the cycle is repeated until the similarity reaches the preset target similarity value.

[0069] Embodiment 6: This embodiment is based on the above embodiment. In step S6, an anomaly detection algorithm is used in the contract approval process to identify potential errors and defects in the draft contract, and automatic correction suggestions are provided through rule matching. An automatic feedback mechanism is introduced to feed back the automatic correction suggestions to the draft contract for correction to obtain a formal contract.

[0070] Embodiment 7, based on the above embodiment, in step S7, after the contract is approved, the digital signature process is executed, the contract document is digitally signed by a digital signature and encryption algorithm to obtain a signed contract;

[0071] Embodiment 8: This embodiment is based on the above embodiment. In step S8, the automatic effectiveness mechanism is referenced, and the contract is marked as effective after being signed and archived in the contract library.

[0072] Embodiment 9, based on the above embodiment, in step S3, special terms are identified by a classification algorithm, specifically including the following steps:

[0073] Step S31: Setting classification targets, classifying contract terms, and setting classification labels for contract terms, including compensation terms, breach of contract terms, and term terms;

[0074] Step S32: preparing training data, preparing three training data sets for three classification labels from existing contract terms, each data set containing sample data;

[0075] Step S33: introducing the SVM model, extracting text features from the draft contract, and converting it into a vector form;

[0076] Step S34: weighted processing, assigning weights according to the classification labels of the contract terms, and introducing weighted nearest neighbor feature selection to perform weighted processing on the sample data;

[0077] Step S35: Model training, screening out important features in the contract terms through forward greedy search, and starting to train the SVM model;

[0078] Step S36: Model optimization, using the alternating direction multiplier method to optimize the SVM model parameters;

[0079] Step S37: Identification and classification: input the pre-processed draft contract terms into the trained SR-SVM model, classify them according to the characteristics of the terms, and identify the classification labels of the contract terms;

[0080] Embodiment 10, based on the above embodiment, step S34 specifically includes the following contents:

[0081] Calculate neighborhood distance: For each sample data, select k nearest neighbor samples and calculate the neighborhood distance respectively;

[0082] Calculate standard deviation: Calculate the standard deviation of the class label for each sample as a measure of sample quality;

[0083] Assign weights: Weight the samples according to the standard deviation of the class labels of each sample. Samples with high class label standard deviations have lower weights, and samples with low class label standard deviations have higher weights.

[0084] Example 11, see Figure 2 This embodiment is based on the above embodiment. In step S35, important features in the contract terms are screened out through forward greedy search, which specifically includes the following steps:

[0085] Step S351: define the initial state, define the initial state of the root node as the set of all contract terms, each contract term corresponds to a feature, each root node represents a feature set, and a subset of the feature set is a feature combination;

[0086] Step S352: introducing an evaluation function, and using an evaluation function based on information gain to calculate the importance of each feature in the classification of contract terms;

[0087] Step S353: Select feature combinations, use the evaluation function to evaluate all feature combinations in the root node, select m feature combinations with the highest evaluation function values ​​as the child nodes of the second layer, and for each child node of the second layer, use the evaluation function again to select m optimal child nodes to generate m×m child nodes of the third layer;

[0088] Step S354: iterative operation, setting the depth d of the tree, iterating according to step S353, generating m×m child nodes based on the optimal child node in each iteration, until the depth d of the tree is reached and the iteration is terminated, generating a feature selection tree, each leaf node of the feature selection tree contains a feature combination;

[0089] Step S355: Introduce a greedy algorithm, and use the greedy algorithm to generate the current optimal feature combination for each leaf node of the feature selection tree. Starting from the feature combination of the leaf node, combined with the feature set of the upper node, the current optimal feature combination is selected layer by layer;

[0090] Step S356: Select the optimal feature set, repeat the iteration step S355, record the current best feature combination obtained in each iteration, select the best feature combination layer by layer, and obtain the optimal feature set;

[0091] Step S357: Output the result, outputting the optimal feature set as the important features in the contract terms, including the combination of different types of contract terms;

[0092] In this embodiment, the contract terms features include compensation amount, performance period, breach of contract clause, service quality standard and payment method. The importance of each feature is calculated through the evaluation function, and the three feature combinations with the highest information gain are selected as the second-layer nodes.

[0093] Second-level subnode 1: compensation amount and breach of contract clause;

[0094] Second-layer subnode 2: performance period and service quality standards;

[0095] Second-layer subnode 3: payment method and compensation amount;

[0096] The optimal feature combination is selected layer by layer through a greedy algorithm, and finally "compensation amount" and "breach of contract clause" are output as the optimal feature combination.

[0097] Embodiment 12, based on the above embodiment, step S352 specifically includes the following contents: selecting a feature combination with the highest information gain based on an evaluation function of information gain, the formula of the evaluation function is as follows: ;

[0098] in, is the entropy of the set of all contract terms, Indicates the classification of contract terms The probability of ;

[0099] in, is the set of feature A in all contract terms The information gain on Let the value of feature A in S be When the subset is the total number of features, for The entropy of .

[0100] In this example, we have 100 contract clauses, which are classified into three categories: compensation clauses, breach of contract clauses, and term clauses. The probability of classification is is 40%, 35% and 25%, calculate , we have a feature A, which is whether the clause contains the word "breach of contract". The value of feature A is yes and no, and the set of contract clauses with "A=yes" A set of contract clauses containing 60 contracts where "A = No" There are 40 contracts, and the information gain is calculated to be 0.35, so the information gain of feature A for classifying contract terms is 0.35.

[0101] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0102] 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 variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

[0103] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. A fully automated circulation and validation method for smart contracts, characterized by: The method comprises the following steps: Step S1: Contract information entry; Step S2: Verify contract items and perform anomaly detection; Step S3: Automatically generate a draft contract; Step S4: Check and adjust contract information; Step S5: Automatically approve the contract; Step S6: Perform exception handling and introduce automatic feedback mechanism; Step S7: Automatically sign the contract; Step S8: Automatic processing of contract effectiveness; Step S9: Dynamic contract management.

2. A fully automated smart contract circulation and validation method according to claim 1, characterized in that: In step S1, a contract draft is defined, and the winning bid information is obtained from the bidding results and the winning bid notice. The winning bid information includes the winning bid items and the contract parties. The winning bid information is preprocessed and formatted using a data preprocessing algorithm to obtain the winning bid data.

3. According to claim 1, a fully automated smart contract circulation and validation method is characterized by: In step S2, a set of contract templates is predefined, including contract items and corresponding formats. A data verification algorithm is used to check whether the winning items are consistent with the contract items through an item comparison mechanism. If inconsistencies occur, they are marked and a difference report is generated.

4. According to claim 1, a fully automated circulation and validation method for smart contracts is characterized by: In step S3, a template generation algorithm is used to fill in the contract template according to the bid winning data to obtain a contract draft. For special contracts, special terms are identified through a classification algorithm, and the contract template is adjusted before filling in.

5. According to claim 1, a fully automated circulation and validation method for smart contracts is characterized by: In step S4, the draft contract is checked using a text comparison algorithm. If any inconsistency is found, a comparison report is generated and the draft contract is modified based on the comparison report.

6. A fully automated smart contract circulation and validation method according to claim 1, characterized in that: In step S6, an anomaly detection algorithm is used in the contract approval process to identify potential errors and defects in the draft contract, and automatic correction suggestions are provided through rule matching. An automatic feedback mechanism is introduced to feed back the automatic correction suggestions to the draft contract for correction to obtain a formal contract.

7. A fully automated smart contract circulation and validation method according to claim 1, characterized in that: In step S7, after the contract is approved, the digital signature process is executed, and the contract document is digitally signed through a digital signature and encryption algorithm to obtain a signed contract.

8. According to claim 1, a fully automated smart contract circulation and validation method is characterized by: In step S8, the automatic effectiveness mechanism is referenced, and the contract is marked as effective after signing and archived in the contract library.

9. A fully automated smart contract circulation and validation method according to claim 4, characterized in that: In step S3, special terms are identified through a classification algorithm, which specifically includes the following steps: Step S31: Setting classification targets, classifying contract terms, and setting classification labels for contract terms, including compensation terms, breach of contract terms, and term terms; Step S32: preparing training data, preparing three training data sets for three classification labels from existing contract terms, each data set containing sample data; Step S33: introducing the SVM model, extracting text features from the draft contract, and converting it into a vector form; Step S34: weighted processing, assigning weights according to the classification labels of the contract terms, and introducing weighted nearest neighbor feature selection to perform weighted processing on the sample data; Step S35: Model training, screening out important features in the contract terms through forward greedy search, and starting to train the SVM model; Step S36: Model optimization, using the alternating direction multiplier method to optimize the SVM model parameters; Step S37: Identification and classification: input the pre-processed draft contract terms into the trained SR-SVM model, classify them according to the characteristics of the terms, and identify the classification labels of the contract terms.

10. A fully automated smart contract circulation and validation method according to claim 9, characterized in that: In step S35, important features in the contract terms are screened out through forward greedy search, which specifically includes the following steps: Step S351: define the initial state, define the initial state of the root node as the set of all contract terms, each contract term corresponds to a feature, each root node represents a feature set, and a subset of the feature set is a feature combination; Step S352: introducing an evaluation function, and using an evaluation function based on information gain to calculate the importance of each feature in the classification of contract terms; Step S353: Select feature combinations, use the evaluation function to evaluate all feature combinations in the root node, select m feature combinations with the highest evaluation function values ​​as the child nodes of the second layer, and for each child node of the second layer, use the evaluation function again to select m optimal child nodes to generate m×m child nodes of the third layer; Step S354: iterative operation, setting the depth d of the tree, iterating according to step S353, generating m×m child nodes based on the optimal child node in each iteration, until the depth d of the tree is reached and the iteration is terminated, generating a feature selection tree, each leaf node of the feature selection tree contains a feature combination; Step S355: Introduce a greedy algorithm, and use the greedy algorithm to generate the current optimal feature combination for each leaf node of the feature selection tree. Starting from the feature combination of the leaf node, combined with the feature set of the upper node, the current optimal feature combination is selected layer by layer; Step S356: Select the optimal feature set, repeat the iteration step S355, record the current best feature combination obtained in each iteration, select the best feature combination layer by layer, and obtain the optimal feature set; Step S357: Output the result, output the optimal feature set as the important features in the contract terms, including the combination of different types of contract terms.

Images