A contract intelligent review method based on natural language understanding

Abstract

The present invention relates to the field of document review technology and discloses a method for intelligent contract review based on natural language understanding. The method comprises the following steps: S1, constructing a contract review knowledge graph: an artificially constructed label system and review points together constitute a contract review knowledge graph; S2, implementing review rules; an algorithm engineer implements specific contract review functions based on the knowledge graph; S3, intelligent contract review. The method for intelligent contract review based on natural language understanding sorts out contract review requirements and forms a contract review knowledge graph containing labels, review points, review items, review rules, risk warning information, etc. The algorithm model is trained based on the contract review knowledge graph, review rules are implemented, and a review checklist can be generated based on the content of the instruction graph. Based on the review checklist and review position, the method intelligently schedules relevant algorithm models and review modules to ultimately obtain the contract review results.

Description

Technical Field

[0001] The present invention belongs to the technical field of document review, and specifically relates to an intelligent contract review method based on natural language understanding. Background Art

[0002] A contract is the contractual carrier of business cooperation between enterprises. Contract review is an indispensable part of the daily operations of corporate entities. Traditional contract review methods rely heavily on manual review operations, which has long occupied a large amount of corporate legal human resources. At the same time, because each contract reviewer has different experience, understanding of laws and regulations, and control of risk points, the final contract review standards and results are also different, which invisibly increases the risks of corporate compliance and contract dispute litigation. An intelligent contract review system based on natural language understanding can use fixed and unified standards for various contract risks in accordance with the requirements of laws and regulations and the needs of corporate operations, and conduct automated AI reviews. This can not only reduce the workload of corporate legal affairs, but also improve the standards and norms of contract reviews.

[0003] Smart contract review, as an emerging functional service, has appeared in some paperless office services or document management services. However, these contract reviews mainly compare the current contract with the template contract and judge whether there are risks based on the differences in the text. This is very mechanical and has no obvious effect on reducing the workload of legal affairs. It even adds some steps to the contract review and cannot truly solve the problems of cumbersome work, inconsistent standards, and manpower consumption in contract review. Summary of the Invention

[0004] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a contract intelligent review method based on natural language understanding, which effectively solves the problems raised in the above background technology.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a contract intelligent review method based on natural language understanding, the review method comprising the following steps:

[0006] S1. Constructing a contract review knowledge graph: The manually constructed labeling system and review points together constitute the contract review knowledge graph;

[0007] S2. Implement review rules: Algorithm engineers implement specific contract review functions based on the knowledge graph;

[0008] S3. Smart Contract Review: Before reviewing a contract, users can customize their own review checklists. All review points for each contract type are combined into a default review checklist for that contract type. Users can filter out the review points they need from the default review checklist and edit the review point name, sub-review item name, and risk warning information of the review point to ultimately form the review checklist they need. The subsequent smart review function will automatically perform different review operations based on the different review points involved in each review checklist.

[0009] Preferably, the construction of the contract review knowledge graph in S1 includes the following steps:

[0010] S1-1. Sort out contract review requirements and consolidate relatively abstract contract review requirements into a limited number of review points. For example, a contract for the sale of goods may have a review point for "contract amount review";

[0011] S1-2. Consolidate the various risks included in the review points into a limited number of sub-review items. Each review item reviews a specific risk point. For example, the review point "Contract Amount Review" will have several sub-review items such as "Amount Capitalization Consistency" and "Total Amount Accuracy";

[0012] S1-3: Based on the review points and their sub-review items, abstract the labels that may be needed to implement the review logic and determine the label connotations. Then, randomly select some real contracts and conduct trial labeling of the above labels to confirm whether the above labels are applicable and their connotations are accurate in real contracts.

[0013] S1-4. If, during the trial labeling process, it is found that some tags are set unreasonably or their connotations are inaccurate, repeat the above two steps, and iterate the tags and their connotations until the algorithm is usable and the connotations are clear. Integrate the above tags into a tag system with a certain hierarchical relationship. For example, a goods sales contract may have a "payment" clause and a "breach of contract" clause, and under the "payment" clause there are clauses such as "total price of goods" and "delivery time". These clauses should be organized into a set of hierarchical tags;

[0014] S1-5. Label a series of contract text data based on the constructed labeling system and labeling connotations, and synchronize the labeling system and labeled data with the algorithm engineer for training the algorithm model;

[0015] S1-6. Based on the labeling system, specific review rules are established for each sub-item of the review point. For example, "If the total contract amount in the 'Total Price of Goods' label does not contain both uppercase and lowercase amounts and currency types, there is a risk."

[0016] S1-7. All review points constitute the contract review knowledge graph, which includes all tags, review points, review items, review rules, risk warning information, etc.

[0017] S1-8. Use the contract knowledge graph synchronization module to synchronize the sorted knowledge graph to the algorithm engineer to implement the review points and review rules. At the same time, according to the actual effect of the contract intelligent review, adjust the relevant tags / review rules and other knowledge graph information to optimize and iterate the contract review knowledge graph.

[0018] Preferably, the review rules implemented in S2 include the following steps:

[0019] S2-1. Use the contract knowledge graph synchronization module to extract information related to the label system. Based on the label system, generate model training data by annotating the data according to the label hierarchy structure;

[0020] S2-2. Based on the above labeling system and training data, train the algorithm model so that the algorithm can accurately identify various labels and elements in such contracts based on the manually constructed labeling system;

[0021] S2-3. Use the contract knowledge graph synchronization module to extract review points, sub-review items, review rules, and related label information. Leverage model labels and elements to implement review rules for all sub-review items at each review point, and encapsulate this review point into a callable review module. For example, in the "Contract Total Amount Review" review point, use the label model to obtain the clause containing the contract total amount. Then, use the element extraction model to extract the specific total amount element from this clause. Use the total amount element to further implement subsequent review rules. After encapsulating this review rule into a review module, this review point can be called at any time to review risks related to the "Contract Total Amount."

[0022] S2-4. Use real contracts to test the effectiveness of the smart contract review algorithm. If the review results are incorrect, fine-tune the review rules under the erroneous review items and synchronize the fine-tuning results to the contract review knowledge graph. Ultimately, iterate to find the most effective smart contract review algorithm.

[0023] S2-5. Use the contract knowledge graph synchronization module to extract review points, sub-review items, risk warning text, and related label information to generate a review list.

[0024] Preferably, the smart contract review in S3 includes the following steps:

[0025] S3-1. The user uploads the contract and begins reviewing it. Meanwhile, the document parsing module parses the contract in formats such as docx and PDF into contract text information in a unified format.

[0026] S3-2. The feature extraction module will first identify basic information in the contract, such as the contract type and information about the contracting parties, and display this information to the user, assisting the user in completing the necessary information for intelligent review and selecting the corresponding review checklist. For example, the review points for "shop rental" contracts and "goods sales" contracts are different. In the "goods sales" contract, the review focus also varies depending on whether the user is a "buyer" or a "seller." The feature extraction module can help the user quickly select the necessary information for contract review and select the review checklist to proceed to the subsequent review steps;

[0027] S3-3. The user independently selects a review stance and a list of review points, and proceeds with the subsequent intelligent review steps. The algorithm scheduling module then selects the review modules to be executed based on the review list and review stance, and combines these modules into a review module execution tree required for this review. The algorithm scheduling module also obtains all the clause classification models and element extraction models required by these review modules, and generates an algorithm model execution tree required for this review based on the hierarchical relationships between all tags and elements described in the tag system. Following the algorithm model execution tree and the review module execution tree, all models and review modules are called sequentially.

[0028] S3-4, the review module, obtains the clause and element information required for review from the execution results of the clause classification model and element extraction model. Based on the information in the contract review knowledge graph, it sequentially executes all sub-review items, determines whether there are risks based on the review rules of the sub-review items, and generates and returns all review results.

[0029] S3-5. The results returned by the review module are used to generate risk cards with specific risk warning information based on the configuration of the review checklist. The user can then choose to ignore the risk or annotate the relevant risk based on the prompt information on the risk card. The risk information that is not ignored and the annotation information will be saved on the contract review platform. The contract can also be exported, and the exported contract can also save the risk warning information and user annotation information in the form of annotations;

[0030] S3-6. After a user modifies a contract on the contract review platform, or modifies a contract offline and re-uploads it, the modified contract will be saved as a new version of the original contract. Each version of the contract can be compared with the original contract using the version comparison module to confirm the modified content of each version of the contract. The new version of the contract can also be intelligently reviewed to confirm the risk situation.

[0031] S3-7. After confirming that all risk information of the contract has been resolved or all risks can be ignored, the contract signing process can begin.

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

[0033] 1. Streamline the legal knowledge involved in contract review, solidify the contract review knowledge graph, standardize contract review methods, and train algorithmic models to understand contract content and implement review rules, thereby enabling intelligent contract review, further improving the efficiency and standardization of contract review. Furthermore, it can be customized to meet specific contract review needs, using machines to replicate the thought process and review process of legal personnel when reviewing contracts to the greatest extent possible.

[0034] 2. To meet different review needs, users can select different review points, review checklists, sub-review items, and contract holders, and implement different review rules for different review scenarios. Users can also edit the names of review points, risk warning information, and other content, making smart contract review more closely aligned with users' actual review needs.

[0035] 3. The review algorithm implementation process and the contract review knowledge graph construction process. While the two processes iteratively evolve and improve the effects within their own processes, the output results can assist in the optimization and iteration of the other process, so that the two processes can ultimately achieve the best results more efficiently. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0037] Figure 1 It is a schematic diagram of the process of the present invention;

[0038] Figure 2 This is a schematic diagram of the intelligent contract review process of the present invention;

[0039] Figure 3 A flowchart illustrating the implementation of review rules in the present invention. DETAILED DESCRIPTION

[0040] 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.

[0041] Embodiment 1, by Figure 1 The present invention includes a contract intelligent review method based on natural language understanding, and the review method includes the following steps:

[0042] S1. Constructing a contract review knowledge graph: The manually constructed labeling system and review points together constitute the contract review knowledge graph;

[0043] S2. Implement review rules: Algorithm engineers implement specific contract review functions based on the knowledge graph;

[0044] S3. Smart Contract Review: Before reviewing a contract, users can customize their own review checklists. All review points for each contract type are combined into a default review checklist for that contract type. Users can filter out the review points they need from the default review checklist and edit the review point name, sub-review item name, and risk warning information of the review point to ultimately form the review checklist they need. The subsequent smart review function will automatically perform different review operations based on the different review points involved in each review checklist.

[0045] In Example 2, the construction of the contract review knowledge graph in S1 includes the following steps:

[0046] S1-1. Sort out contract review requirements and consolidate relatively abstract contract review requirements into a limited number of review points. For example, a contract for the sale of goods may have a review point for "contract amount review";

[0047] S1-2. Consolidate the various risks included in the review points into a limited number of sub-review items. Each review item reviews a specific risk point. For example, the review point "Contract Amount Review" will have several sub-review items such as "Amount Capitalization Consistency" and "Total Amount Accuracy";

[0048] S1-3: Based on the review points and their sub-review items, abstract the labels that may be needed to implement the review logic and determine the label connotations. Then, randomly select some real contracts and conduct trial labeling of the above labels to confirm whether the above labels are applicable and their connotations are accurate in real contracts.

[0049] S1-4. If, during the trial labeling process, it is found that some tags are set unreasonably or their connotations are inaccurate, repeat the above two steps, and iterate the tags and their connotations until the algorithm is usable and the connotations are clear. Integrate the above tags into a tag system with a certain hierarchical relationship. For example, a goods sales contract may have a "payment" clause and a "breach of contract" clause, and under the "payment" clause there are clauses such as "total price of goods" and "delivery time". These clauses should be organized into a set of hierarchical tags;

[0050] S1-5. Label a series of contract text data based on the constructed labeling system and labeling connotations, and synchronize the labeling system and labeled data with the algorithm engineer for training the algorithm model;

[0051] S1-6. Based on the labeling system, specific review rules are established for each sub-item of the review point. For example, "If the total contract amount in the 'Total Price of Goods' label does not contain both uppercase and lowercase amounts and currency types, there is a risk."

[0052] S1-7. All review points constitute the contract review knowledge graph, which includes all tags, review points, review items, review rules, risk warning information, etc.

[0053] S1-8. Use the contract knowledge graph synchronization module to synchronize the sorted knowledge graph to the algorithm engineer to implement the review points and review rules. At the same time, according to the actual effect of the contract intelligent review, adjust the relevant tags / review rules and other knowledge graph information to optimize and iterate the contract review knowledge graph.

[0054] In the third embodiment, the review rules are implemented in S2 in the following steps:

[0055] S2-1. Use the contract knowledge graph synchronization module to extract information related to the label system. Based on the label system, generate model training data by annotating the data according to the label hierarchy structure;

[0056] S2-2. Based on the above labeling system and training data, train the algorithm model so that the algorithm can accurately identify various labels and elements in such contracts based on the manually constructed labeling system;

[0057] S2-3. Use the contract knowledge graph synchronization module to extract review points, sub-review items, review rules, and related label information. Leverage model labels and elements to implement review rules for all sub-review items at each review point, and encapsulate this review point into a callable review module. For example, in the "Contract Total Amount Review" review point, use the label model to obtain the clause containing the contract total amount. Then, use the element extraction model to extract the specific total amount element from this clause. Use the total amount element to further implement subsequent review rules. After encapsulating this review rule into a review module, this review point can be called at any time to review risks related to the "Contract Total Amount."

[0058] S2-4. Use real contracts to test the effectiveness of the smart contract review algorithm. If the review results are incorrect, fine-tune the review rules under the erroneous review items and synchronize the fine-tuning results to the contract review knowledge graph. Ultimately, iterate to find the most effective smart contract review algorithm.

[0059] S2-5. Use the contract knowledge graph synchronization module to extract review points, sub-review items, risk warning text, and related label information to generate a review list.

[0060] In the fourth embodiment, the smart contract review in S3 includes the following steps:

[0061] S3-1. The user uploads the contract and begins reviewing it. Meanwhile, the document parsing module parses the contract in formats such as docx and PDF into contract text information in a unified format.

[0062] S3-2. The feature extraction module will first identify basic information in the contract, such as the contract type and information about the contracting parties, and display this information to the user, assisting the user in completing the necessary information for intelligent review and selecting the corresponding review checklist. For example, the review points for "shop rental" contracts and "goods sales" contracts are different. In the "goods sales" contract, the review focus also varies depending on whether the user is a "buyer" or a "seller." The feature extraction module can help the user quickly select the necessary information for contract review and select the review checklist to proceed to the subsequent review steps;

[0063] S3-3. The user independently selects a review stance and a list of review points, and proceeds with the subsequent intelligent review steps. The algorithm scheduling module then selects the review modules to be executed based on the review list and review stance, and combines these modules into a review module execution tree required for this review. The algorithm scheduling module also obtains all the clause classification models and element extraction models required by these review modules, and generates an algorithm model execution tree required for this review based on the hierarchical relationships between all tags and elements described in the tag system. Following the algorithm model execution tree and the review module execution tree, all models and review modules are called sequentially.

[0064] S3-4, the review module, obtains the clause and element information required for review from the execution results of the clause classification model and element extraction model. Based on the information in the contract review knowledge graph, it sequentially executes all sub-review items, determines whether there are risks based on the review rules of the sub-review items, and generates and returns all review results.

[0065] S3-5. The results returned by the review module are used to generate risk cards with specific risk warning information based on the configuration of the review checklist. The user can then choose to ignore the risk or annotate the relevant risk based on the prompt information on the risk card. The risk information that is not ignored and the annotation information will be saved on the contract review platform. The contract can also be exported, and the exported contract can also save the risk warning information and user annotation information in the form of annotations;

[0066] S3-6. After a user modifies a contract on the contract review platform, or modifies a contract offline and re-uploads it, the modified contract will be saved as a new version of the original contract. Each version of the contract can be compared with the original contract using the version comparison module to confirm the modified content of each version of the contract. The new version of the contract can also be intelligently reviewed to confirm the risk situation.

[0067] S3-7. After confirming that all risk information of the contract has been resolved or all risks can be ignored, the contract signing process can begin.

[0068] The above tags indicate that a section of the contract text has special meanings, such as "liability for breach of contract", "confidentiality clause", "payment clause", etc.

[0069] The above elements represent some key information in a contract text, such as "contracting party" and "total contract amount".

[0070] The above labeling system refers to the relationship between labels, between labels and elements, and between elements. For example, the "Payment Terms" label is in parallel with the "Breach of Contract Liability" label, while the "Payment Terms" label contains the "Total Price of Goods" label and the "Delivery Time" label. The "Total Price of Goods" label contains the "Total Contract Amount" element.

[0071] The above review points refer to a specific review point in the contract review, such as "review of dispute resolution clauses", which usually includes several review items;

[0072] The above-mentioned review items refer to a key review point within the review points, such as the review item "Risk of missing dispute resolution clauses" within the review point "Review of dispute resolution clauses" and the review item "Risk of concurrent litigation and arbitration." Typically, a review item requires labels, elements, and specific review rules.

[0073] The above review rules refer to the rules for determining certain risks during contract review, such as "there is a risk if the liquidated damages are greater than 30% of the total amount";

[0074] The above review results refer to the final execution results of the review items, including the result name, whether there is risk, risk warning information, etc.

[0075] The above-mentioned review modules refer to services that perform specific review points;

[0076] The above review checklist refers to a collection of all review points for a certain contract type and is used for the review of a specific contract;

[0077] The above review stance refers to the user's review stance when reviewing the contract, such as the lessee, buyer, and seller. Different review stances may result in different review points, review items, and review rules.

[0078] The contract review knowledge graph mentioned above refers to a knowledge graph that contains all tags, elements, review points, and review rules for a certain type of contract.

[0079] The aforementioned clause classification model refers to an algorithmic model that identifies key paragraphs in a contract text and labels them accordingly. For example, it identifies the "liability for breach of contract" clause in a contract and labels it "liability for breach of contract";

[0080] The above-mentioned feature extraction model refers to an algorithmic model that identifies key segments in a paragraph and labels them accordingly. For example, it identifies the "total amount of payment" in the payment terms and labels it as "total amount of payment";

[0081] The aforementioned legal researchers are professionals with legal backgrounds and professional knowledge, responsible for defining labels and elements, and organizing the labeling system, review points, review items, and review rules.

[0082] The above-mentioned algorithm engineers refer to algorithm engineering personnel who are responsible for the implementation of clause classification models, factor extraction models, and review projects.

[0083] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0084] While 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 these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent contract review based on natural language understanding, characterized by: The review method comprises the following steps: S1. Constructing a contract review knowledge graph: The manually constructed labeling system and review points together constitute the contract review knowledge graph; S2. Implement review rules: Algorithm engineers implement specific contract review functions based on the knowledge graph; S3. Smart Contract Review: Before contract review, users customize their own review checklist. All review points for each contract type are combined into a default review checklist for that contract type. Users filter the review points they need from the default review checklist and edit the review point name, sub-review item name, and risk warning information to form their desired review checklist. The subsequent smart review function will automatically perform different review operations based on the different review points involved in each review checklist. The construction of the contract review knowledge graph in S1 includes the following steps: S1-1. Sort out contract review requirements and integrate them into n review points; S1-2. Integrate the various risks included in the review point into m sub-review items, each of which reviews a specific risk point; S1-3. Based on the review points and their sub-review items, extract the tags required to implement the review logic and determine the tag connotations. Then, randomly select a number of real contracts and conduct trial labeling of the tags to confirm whether the tags are applicable and their connotations are accurate in real contracts. S1-4. If, during the trial labeling process, it is found that some labels are set irrationally or their connotations are inaccurate, then repeat the above steps S1-1 and S1-2, and iterate the labels and their connotations until the algorithm is usable and the connotations are clear, and integrate the above labels into a hierarchical label system; S1-5. Label a series of contract text data based on the constructed labeling system and labeling connotations, and synchronize the labeling system and labeled data with the algorithm engineer for training the algorithm model; S1-6. Based on the labeling system, formulate specific review rules for each sub-review item of the review point; S1-7. Build a contract review knowledge graph based on all review points. The graph includes all tags, review points, review items, review rules, and risk warning information. S1-8. Use the contract knowledge graph synchronization module to synchronize the sorted knowledge graph to the algorithm engineer to implement the review points and review rules. At the same time, according to the actual effect of the contract intelligent review, adjust the relevant label / review rule knowledge graph information and optimize the iterative contract review knowledge graph.

2. The intelligent contract review method based on natural language understanding according to claim 1, characterized in that: The implementation of the review rules in S2 includes the following steps: S2-1. Use the contract knowledge graph synchronization module to extract information related to the label system. Based on the label system, generate model training data by annotating the data according to the label hierarchy structure; S2-2. Based on the above labeling system and training data, train the algorithm model so that the algorithm can accurately identify various labels and elements in such contracts based on the manually constructed labeling system; S2-3. Use the contract knowledge graph synchronization module to extract review points, sub-review items, review rules, and related tag information. Leverage model tags and elements to implement review rules for all sub-review items at each review point, and encapsulate this review point into a callable review module. S2-4. Use real contracts to test the effectiveness of the smart contract review algorithm. If the review results are incorrect, fine-tune the review rules under the erroneous review items and synchronize the fine-tuning results to the contract review knowledge graph. Ultimately, iterate to find the most effective smart contract review algorithm. S2-5. Use the contract knowledge graph synchronization module to extract review points, sub-review items, risk warning text, and related label information to generate a review list.

3. The intelligent contract review method based on natural language understanding according to claim 1, characterized in that: The smart contract review in S3 includes the following steps: S3-1. The user uploads the contract and begins reviewing it. Meanwhile, the document parsing module parses the contract in docx or PDF format into contract text information in a unified format. S3-2. The element extraction module will first identify the basic information in the contract, including the contract type and information about the contracting parties. This basic information will be displayed to the user to assist the user in completing the necessary information for intelligent review and selecting the corresponding review checklist. The element extraction module can help the user quickly select the necessary information for contract review and select the review checklist to proceed to the subsequent review steps. S3-3. The user independently selects a review stance and a list of review points, and proceeds with the subsequent intelligent review steps. The algorithm scheduling module then selects the review modules to be executed based on the review list and review stance, and combines these modules into a review module execution tree required for this review. The algorithm scheduling module also obtains all the clause classification models and element extraction models required by these review modules, and generates an algorithm model execution tree required for this review based on the hierarchical relationships between all tags and elements described in the tag system. Following the algorithm model execution tree and the review module execution tree, all models and review modules are called sequentially. S3-4, the review module, obtains the clause and element information required for review from the execution results of the clause classification model and element extraction model. Based on the information in the contract review knowledge graph, it sequentially executes all sub-review items, determines whether there are risks based on the review rules of the sub-review items, and generates and returns all review results. S3-5. The results returned by the review module are used to generate risk cards with specific risk warning information based on the configuration of the review checklist. The user can then choose to ignore the risk or annotate the relevant risk based on the prompt information on the risk card. The risk information that is not ignored and the annotation information will be saved on the contract review platform. Alternatively, the contract can be exported, and the exported contract can also save the risk warning information and user annotation information in the form of annotations. S3-6. After a user modifies a contract on the contract review platform, or modifies a contract offline and re-uploads it, the modified contract will be saved as a new version of the original contract. Each version of the contract can be compared with the original contract using the version comparison module to confirm the modified content of each version of the contract. The new version of the contract can also be intelligently reviewed to confirm the risk situation. S3-7. After confirming that all risk information of the contract has been resolved or all risks can be ignored, start the contract signing process.

Images