JP7506867B1 - Document correction program, document correction device, and document correction method - Google Patents

Abstract

【assignment】
Support flexible modifications to suit individual documents.
SOLUTION
A document correction program for using a sentence generation model to suggest corrections to a document based on an evaluation of the document, the program causing a computer to function as an acquisition unit, an instruction generation unit, and a suggestion unit, the acquisition unit acquiring draft information regarding a draft of the document and evaluation information indicating an evaluation of the draft, the instruction generation unit generating correction instructions regarding the draft based on the draft information and evaluation information and transmitting them to the sentence generation model, and the suggestion unit transmitting correction suggestion information for revising the draft based on the output of the sentence generation model.
[Selected Figure] Figure 1

Description

The present invention relates to a document correction program, a document correction device, and a document correction method.

The evaluation and creation of documents, particularly legal documents such as contracts and regulations, requires a high level of expertise. Against this background, there are known technologies for checking legal documents and technologies for supporting the revision of legal documents based on the checking.

For example, Patent Document 1 describes a technology that accepts input of information on review results related to legal documents and makes corrections. The technology described in Patent Document 1 makes it possible to correct legal documents by using recommended sentences registered in advance and replacing corresponding parts with the recommended sentences.

JP 2020-080132 A

However, the technology described in Patent Document 1 makes corrections by partially replacing the text with pre-registered recommended sentences, so it can only handle content that is anticipated in advance, and there is an issue that it is difficult to flexibly make corrections to suit individual documents. Therefore, the present invention aims to solve the above problems.

To solve the above problem, a document correction program is provided that utilizes a writing generation model to suggest corrections to a document based on an evaluation of the document, and causes a computer to function as an acquisition unit, an instruction generation unit, and a suggestion unit, the acquisition unit acquires draft information regarding a draft of the document and evaluation information indicating an evaluation of the draft, the instruction generation unit generates correction instructions regarding the draft based on the draft information and evaluation information and transmits them to the writing generation model, and the suggestion unit transmits correction suggestion information for correcting the draft based on the output of the writing generation model.

With this configuration, it is possible to use a text generation model to assist in revising the draft based on the evaluation of the document draft. This makes it possible to flexibly suggest revisions according to the contents of the document, making it easier to revise documents. In particular, even when revising documents that require expertise, such as legal documents, it becomes easier for users to make revisions themselves using the text generation model, making tasks such as creating and reviewing contracts more efficient.

In a preferred embodiment of the present invention, the document is a legal document including a plurality of clauses, and the evaluation information includes the results of evaluating the draft with respect to each clause and/or the relationship between the plurality of clauses.

With this configuration, it is possible to suggest revisions to a document written by clause based on an evaluation of the content of each clause or the relationships between multiple clauses, thereby supporting more appropriate revisions. For example, if a set of regulations is stipulated in multiple clauses, those clauses can be treated as a group and the relationships between them can be confirmed, making it possible to appropriately evaluate the regulations that span multiple clauses. Furthermore, by sending an instruction to revise the draft to the text generation model using that evaluation as evaluation information, it is possible to appropriately suggest revisions for such a group of multiple clauses.

In a preferred embodiment of the present invention, the evaluation information includes an evaluation of any of unnecessary clauses, missing information, and the contents of clauses.

This structure makes it possible to propose appropriate revisions based on an evaluation of whether a clause should be deleted, added, or the content of a clause should be modified.

In a preferred embodiment of the present invention, the command generation unit generates, as the modification command, a sentence that describes instructions regarding the modification for each clause or for each group of multiple clauses that defines a set of regulations, and transmits the generated sentence to the sentence generation model.

With this configuration, it becomes possible to appropriately send amendment commands for legal documents divided into clauses for each clause or for each group of clauses that defines a set of regulations. This makes it easier to execute amendments appropriate for legal documents that include multiple clauses.

In a preferred embodiment of the present invention, the evaluation information includes a risk level for each of the issues raised about the draft, and the instruction generation unit generates, as the correction instructions, instructions to correct the draft for issues that are at or above a predetermined risk level, and transmits the correction instructions to the text generation model.

This configuration makes it possible to suggest modifications to high-risk parts. This makes it possible to propose modifications separately for low-risk parts and parts for which the user should consider whether or not they need to be modified, depending on the risk level.

In a preferred embodiment of the present invention, the evaluation information includes an evaluation statement that describes an evaluation of the draft in natural language, and the instruction generation unit generates a sentence based on the draft, the evaluation statement, and a prepared instruction statement as the correction command and transmits the correction command to the sentence generation model.

This configuration makes it possible to generate sentences in natural language as correction commands, so that correction suggestions can be made, for example, using a conversational sentence generation model that outputs sentences in response to input sentences.

In a preferred embodiment of the present invention, the computer is further made to function as a reevaluation unit, which obtains a proposed revision of the draft as an output of the sentence generation model, obtains reevaluation information indicating an evaluation of the proposed revision, determines whether the evaluation satisfies a preset criterion, and if the criterion is satisfied, causes the suggestion unit to transmit the proposed revision information, and if the criterion is not satisfied, causes the instruction generation unit to generate a revision instruction for the proposed revision based on the revision information and reevaluation information based on the proposed revision and transmit the revision instruction to the sentence generation model.

This configuration prevents proposed revisions that still have problems from being sent as is, and if there are problems, the proposal can be reevaluated and revised again. This is expected to improve the quality of proposed revisions.

In a preferred embodiment of the present invention, the re-evaluation unit causes the suggestion unit and the instruction generation unit to repeatedly obtain re-evaluation information and send revision instructions until the evaluation of the revision proposal satisfies a preset criterion.

This configuration allows the evaluation and correction to be repeated until the criteria are met. This further reduces the occurrence of problems where incomplete parts remain after correction, and allows for higher quality correction suggestions.

In a preferred embodiment of the present invention, the reevaluation information includes a risk level for each of the issues raised in the proposed revision, and the criteria include that the risk levels of all of the issues raised in the proposed revision are equal to or lower than a predetermined risk level.

By configuring in this way, corrections can be repeated until there are no issues that exceed a certain risk level, further improving the quality of correction suggestions.

In a preferred embodiment of the present invention, when a preset termination condition is met, the reevaluation unit terminates the repeated reevaluation and causes the suggestion unit to transmit the proposed revision information, even if the evaluation of the proposed revision does not satisfy a criterion.

By configuring it this way, infinite loops and extremely repetitive processing can be avoided, and the processing load is expected to be reduced.

In a preferred embodiment of the present invention, the command generation unit receives input of a request for correction from a user, generates the correction command based on the request, and transmits the correction command to the sentence generation model.

This configuration makes it possible to propose modifications that reflect the user's wishes. For example, when making a contract with a specific party, there may be multiple possible modification patterns, such as making aggressive modifications (to the party's advantage) or establishing an equal relationship. By accepting the user's wishes and generating a modification command, even in such cases, it is possible to make modification proposals that are more in line with the user's wishes.

In a preferred embodiment of the present invention, the document is a legal document relating to an agreement with a specific party, and the command generation unit acquires background information indicating at least one of the relationship with the party, the type of the agreement, and the wishes of the parties regarding the agreement, generates the correction command based on the background information, and transmits it to the text generation model.

With this configuration, it becomes possible to make more appropriate revision proposals based on the relationship with the other party, the type of arrangement, the requests of the parties, etc. For example, by specifying a contract type as the type of arrangement, it is possible to propose more appropriate revisions that take the contract type into consideration. Furthermore, for example, when entering into a confidentiality agreement, it is possible to propose revisions that are advantageous to the user by selecting whether oneself is the discloser or recipient of information as the relationship with the other party. Furthermore, by selecting, as the request of the parties, whether to be forceful toward the other party (to make the content advantageous to oneself) or to make the content equal, it is possible to make revision proposals that reflect the requests.

In a preferred embodiment of the present invention, the draft information is information about a draft generated by a model.

This configuration allows users to complete the document creation process, from drafting to revising, without having to do it themselves. This allows users to easily create legal documents that require expertise, such as contracts.

In a preferred embodiment of the present invention, the document is a legal document relating to an agreement with a particular party, and the draft information is information about a draft generated by the model based on prior information indicating at least one of the relationship with the party, the type of the agreement, and the parties' wishes regarding the agreement.

This structure allows you to create a draft based on the relationship with the other party, the type of agreement, and the requests of the parties involved. This is expected to improve the quality of the final document.

The present invention can support flexible revisions tailored to individual documents.

1 is a diagram showing a configuration of a document correction system according to an embodiment of the present invention. FIG. 2 is a diagram illustrating an example of a hardware configuration of the document correction system according to the embodiment of the present invention. FIG. 2 is a diagram illustrating a functional configuration of the document correction system according to the embodiment of the present invention. FIG. 4 is a sequence diagram illustrating an example of a processing procedure of the document correction system according to the embodiment of the present invention. 13 is a display example of an evaluation screen of the document correction system according to the embodiment of the present invention. 10 is a flowchart showing a processing procedure related to a reevaluation process of the document revision system according to the embodiment of the present invention 13 is a display example of a correction suggestion screen of the document correction system according to the embodiment of the present invention.

The document correction system of the present invention will be described below with reference to the drawings. The drawings show a preferred embodiment. However, the present invention can be embodied in many different forms and is not limited to the embodiment described in this specification.

For example, in this embodiment, the configuration, operation, etc. of the document correction system are described, but similar effects can be achieved by a document correction device, a document correction system, or a method executed by a document correction device, a document correction program, etc. In this embodiment, the document correction system takes the form of so-called cloud computing, in which the document correction program is executed in a server (document correction device) connected to the client terminal via a network in order to realize its functions in a client terminal (user terminal device).

The document correction program may be provided to a computer device as a computer-readable non-transient storage medium to cause any computer device to function as a document correction device, or may be provided downloadably from an external server. The document correction program may also be an add-on (plug-in) program for document editing software such as Microsoft Word (registered trademark).

In this embodiment, the term "part" may include, for example, a combination of hardware resources implemented by a circuit in the broad sense and software information processing that can be specifically realized by these hardware resources. In this embodiment, "information" is represented, for example, by the physical value of a signal value representing voltage or current, the high or low signal value as a binary bit collection consisting of 0 or 1, or a quantum superposition (so-called quantum bit), and communication and calculations can be performed on a circuit in the broad sense. A circuit in the broad sense is a type of circuitry that is realized by appropriately combining a circuit, a processor, a memory, etc. For example, it is a circuit that includes any of the following: a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), an LSI (Large Scale Integration), an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), etc.

<Definition>
First, definitions of terms will be described. The present invention is a technology for supporting document correction by utilizing a sentence generation model. A sentence generation model refers to any model capable of outputting a sentence. As a sentence generation model, a model generated by machine learning can be typically used. In this embodiment, a case will be described in which a conversational sentence generation model is used, which receives a sentence and generates a sentence as a response to the input sentence.

More specifically, the sentence generation model is a machine learning model for natural language processing, and is a model that has a function of generating sentences based on input information. For example, the sentence generation model is a machine learning model based on a Transformer model having a self-attention mechanism. The Transformer model is, for example, a GPT (Generative Pre-trained Transformer) model. The sentence generation model performs learning using a large data set, making it possible to perform not only general natural language generation tasks but also highly specialized sentence generation tasks. For example, the sentence generation model may be a model trained using a large data set using the GPT-3 or GPT-4 algorithm developed by OpenAI (for example, a ChatGPT model generated by combining learning using a large data set and reinforcement learning using a reward prediction model). In addition, in this embodiment, an example of text information is described as the input to the model, but image information or audio information may also be input.

In this embodiment, the document is assumed to be a legal document such as a contract, rules, or regulations. Some legal documents include units divided into articles, clauses, and items, and in this embodiment, each of these units is called a "clause." In this embodiment, the legal document is a contract. Examples of contracts include documents that exert contractual effects, such as non-disclosure agreements, sales contracts, and real estate lease agreements. Furthermore, this embodiment is not limited to contracts, and the legal document may be, for example, rules, regulations, notices, and the like. Examples of rules include documents of legal norms such as work regulations. Examples of notices include documents declaring policies such as privacy. In addition, legal documents include documents that are somehow related to the law, such as notices and powers of attorney.

In the following, a draft refers to a rough draft of a document. In the present invention, information related to a draft is called draft information, and the draft information includes information indicating the content of sentences that make up at least a part of the draft. For example, the draft information may include only information on clauses to be revised or clauses related to the revised parts, or the draft information may include information on the entire text of the draft.

In this embodiment, a revision command is sent to the sentence generation model to output a revision proposal for the draft, which is then used to make a proposal for revising the draft. The revision proposal may be for the entire text of the draft, or may be for, for example, a portion of the text that is the subject of revision.

A correction command is information indicating a command to cause the writing generation model to output information regarding the correction of the draft. In this embodiment, a conversational writing generation model that takes a sentence as input is used, so a sentence or sentence in natural language is generated and sent as a correction command. Then, the sentence or sentence output by the writing generation model as an answer is received, and a correction suggestion is made.

<System Configuration>
Next, the configuration of the document correction system of this embodiment will be described. FIG. 1 is a diagram showing the configuration of the document correction system of this embodiment. As shown in FIG. 1, the document correction system 0 includes a document correction device 1 and a user terminal device 2, and the document correction device 1 is configured to be able to communicate with a text generation device 3 via a communication network NW. The document correction device 1 and the user terminal device 2 are also configured to be able to communicate with each other via the communication network NW. The document correction device 1 operates as a server, and the user terminal device 2 operates as a client terminal. The text generation device 3 is an external server device that provides a text generation model. Note that the document correction device 1 may be configured to store a text generation model and to be able to correct a document without using an external device, or a part or all of the processing executed by the document correction device 1 in this embodiment may be executed in the user terminal device 2.

In this embodiment, the communication network NW is an IP (Internet Protocol) network, but there are no restrictions on the type of communication protocol, and there are also no restrictions on the type or scale of the network.

It is possible to use a general-purpose server computer or a personal computer as the document correction device 1. It is also possible to configure the document correction device 1 using multiple computers.

The user terminal device 2 may be a smartphone, a tablet terminal, a personal computer, a wearable device, or the like. The user terminal device 2 has an application (a web browser application or a dedicated application) for making requests to the document correction device 1 and receiving responses, and uses the document correction program executed in the document correction device 1 via this application.

The document correction device 1 transmits data in a prescribed format as various display information to the application of the user terminal device 2 based on a request transmitted from the application of the user terminal device 2. The data in the prescribed format is, for example, data in JSON (Javascript Object Notation) format or XML format. The application of the user terminal device 2 renders the display data based on the received data, and displays information such as evaluation information and correction suggestions on the display of the user terminal device 2.

For example, when executing a display process of the revision suggestion information and other information on the web browser application of the user terminal device 2, the web browser application first sends an HTTP (Hypertext Transfer Protocol) request to the web server and receives an HTTP response including the file requested in the HTTP request. The received file is display information including the revision suggestion information, indicates the contents or instructions of the web page to be displayed, and is described using a markup language such as HTML (Hypertext Markup Language). A script, which is a program executed on the web browser application, may be added to the received file. The web browser application draws (renders) a web page in the display area based on the received file. If a script is added to the file, the browser executes the script and displays the web page according to the execution result.

<Hardware Configuration>
2A is a diagram showing an example of the hardware configuration of the document correction device 1. The document correction device 1 includes a control unit 101, a storage unit 102, and a communication unit 103 as the hardware configuration.

The control unit 101 includes one or more processors such as a CPU, and controls the overall operation and processing of the document correction device 1 by executing the document correction program according to the present invention, the OS, browser software, and other applications.

The storage unit 102 is a hard disk drive (HDD), a read only memory (ROM), a random access memory (RAM), etc., and stores the document correction program according to the present invention and data used when the control unit 101 executes processing based on the program. The control unit 101 executes processing based on the document correction program stored in the storage unit 102, thereby realizing the functional configuration described below.

The communication unit 103 controls communication with the communication network NW, and performs input necessary to operate the document correction device 1 and outputs related to the operation results.

FIG. 2(b) is a diagram showing an example of the hardware configuration of a terminal device 90 (user terminal device 2). Similarly, the terminal device 90 also includes, as its hardware configuration, a control unit 901, a storage unit 902, a communication unit 903, an input unit 904, and an output unit 905.

The control unit 901 of the terminal device 90 includes one or more processors such as a CPU, and controls the overall operation and processing of the terminal device 90. The storage unit 902 of the terminal device 90 is a HDD, ROM, RAM, etc., and stores the above-mentioned applications and data used when the control unit 901 executes processing based on a program. The communication unit 903 of the terminal device 90 controls communication with the network. The input unit 904 of the terminal device 90 is a mouse, keyboard, etc., and inputs operation requests from the user/provider to the control unit 901. The output unit 905 of the terminal device 90 is a display, etc., and displays the results of processing by the control unit 901, etc.

<Functional configuration>
Next, the functional configuration according to this embodiment will be described with reference to Fig. 3. The document correction device 1 includes, as its functional configuration, an acquisition unit 11, a command generation unit 12, an evaluation unit 13, and a proposal unit 14. In this embodiment, the evaluation unit 13 functions as the evaluation unit and re-evaluation unit of the present invention. This is a specific implementation of information processing by software stored in the storage unit 102 using hardware (such as the control unit 101). For example, the control unit 101 reads out the document correction program stored in the storage unit 102, expands it into a temporary storage area (such as a RAM) in the storage unit 102, etc., executes it, and controls the communication unit 103, etc., thereby implementing the functions of the document correction device 1.

The acquisition unit 11 acquires draft information related to the document draft and evaluation information indicating the evaluation of the draft. The draft information is acquired based on the draft information transmitted from the user terminal device 2. Specifically, for example, the user can upload a file containing a document draft and acquire draft information by accepting an operation to specify at least a part of the contents of the draft. Note that if the entire draft is to be revised, the operation to specify the part to be revised may be omitted. Also, for example, the configuration may be such that the user uploads a part of the document draft to be revised and draft information is acquired for that part.

Here, the user can also use the draft created using the model. In this embodiment, a case will be described in which the same sentence generation model used for revising the draft is used as the model for generating the draft. For example, the document revision device 1 receives advance information from the user terminal device 2 and transmits a draft generation command based on that to the model, thereby obtaining the draft generated by the model. In addition, a sentence generation model different from the sentence generation model used for revising the draft may be used as the model for generating the draft.

For example, if the document is a legal document regarding an agreement with a specific party, the prior information may include information such as the relationship with the party, the purpose of the agreement, the type of agreement, the requests of the parties regarding the agreement, the content of the agreement, and a template for the legal document. Here, the agreement refers to a contract, rules, regulations, etc. The relationship with the party refers to the position in the agreement, such as the discloser or recipient of information in a confidentiality agreement. The type of agreement may be the contract type specified in the legal document, atypical contracts, types of regulations, etc., and the requests of the parties may include whether the contract is advantageous to the party or an equal contract, or any other arbitrary request. This prior information may be input in a format in which the user selects from preset options, or may be text information freely written in natural language.

The advance information may also be term sheet information that indicates the contents of a term sheet used in contract negotiations. A term sheet is a table that summarizes the contents of a proposal or the contents agreed upon with the other party regarding the main items of a contract, and term sheet information is information that includes the contents described in the term sheet. The advance information may also be question information based on the questions and question results generated by questioning software that asks questions about the contents to be included in a legal document (e.g., a contract).

The evaluation information is information that indicates the evaluation of the draft. In this embodiment, the evaluation unit 13 of the document correction device 1 performs an evaluation of the draft, and the acquisition unit 11 acquires the result as evaluation information. However, the present invention is not limited to this, and the evaluation information may be acquired from an external device or the user terminal device 2. The evaluation process for the draft and the evaluation information will be described in detail later.

In addition, the acquisition unit 11 of this embodiment further acquires background information about the background of the contract or the like stipulated by the target legal document. The background of the contract or the like is information such as the relationship with the other party, the purpose of the agreement, the type of agreement, the requests of the parties regarding the agreement, the content to be agreed upon, and a template for the legal document. As background information, information similar to the above-mentioned prior information can be acquired. In this embodiment, evaluation and generation of correction commands are performed based on such background information.

That is, for example, if the document is a legal document regarding an arrangement with a specific party, background information would include information such as the relationship with the party, the purpose of the arrangement, the type of arrangement, the requests of the parties regarding the arrangement, the content of the arrangement, and a template for the legal document. Here, an arrangement refers to a contract, rules, regulations, etc. The relationship with the party refers to the position in the arrangement, for example, the discloser or recipient of information in a confidentiality agreement. Examples of the type of arrangement include the contract type specified in the legal document, atypical contracts, and types of regulations or rules, and the requests of the parties refer to wishes such as whether the contract should be advantageous to the party or an equal contract, comments received from the other party to the arrangement, and other arbitrary requests.

These background information may be input in a format in which the user selects from preset options, or may be text information freely written in natural language. The background information may also be generated by a text generation model that receives legal documents and extraction commands to extract background information from the legal documents. For example, the extraction commands are commands that estimate which contract type the input legal document corresponds to from among multiple predetermined contract types.

The command generation unit 12 generates revision commands related to the draft based on the draft information and evaluation information acquired by the acquisition unit 11, and transmits the generated command to the sentence generation model. In this embodiment, input is made to the sentence generation model by transmitting revision commands to the sentence generation device 3 that provides the sentence generation model. In this embodiment, the command generation unit 12 generates revision commands as sentences in natural language using the draft information and evaluation information. The process of generating revision commands will be described in detail later.

The evaluation unit 13 functions as a re-evaluation unit of the present invention, and receives a proposed revision of the draft from the text generation device 3 as the output of the text generation model in response to a revision command. It then obtains re-evaluation information indicating an evaluation of the proposed revision, and determines whether or not the evaluation satisfies a preset criterion. Note that in this embodiment, the evaluation unit 13 (re-evaluation unit) performs an evaluation of the proposed revision, and generates re-evaluation information as a result, but it may also be possible to transmit the proposed revision to an external evaluation device, and obtain re-evaluation information from the evaluation device as the evaluation result.

If the evaluation satisfies the criteria, the evaluation unit 13 (re-evaluation unit) causes the proposal unit 14 to send revision proposal information. On the other hand, if the evaluation does not satisfy the criteria, the evaluation unit 13 instructs the instruction generation unit 12 to generate a revision command for the revision proposal based on the revision information and re-evaluation information for the revision proposal. The instruction generation unit 12 then sends the revision command to the sentence generation model, which in turn generates a revision proposal for the revision proposal.

The evaluation unit 13 (reevaluation unit) of this embodiment is configured to repeat obtaining and evaluating revision proposals until the evaluation of the revision proposals meets a predetermined criterion or a termination condition is met. The predetermined criterion can be, for example, that the risk level for each of the issues mentioned below contained in the reevaluation information is equal to or lower than a predetermined risk level. In addition, the termination condition can be, for example, setting an upper limit on the number of repetitions or a lower limit on the number of characters changed by the revision. The reevaluation process will be described in detail later.

The suggestion unit 14 transmits revision suggestion information based on the output of the sentence generation model in response to the revision command. In this embodiment, in accordance with the judgment result by the evaluation unit 13 (re-evaluation unit), display information including the revision suggestion information is transmitted to the user terminal device 2, whereby the user terminal device 2 displays the revision suggestion.

The suggestion unit 14 of this embodiment generates revision suggestion information including sentences in natural language to assist in revising the draft based on the output of the sentence generation model acquired by the evaluation unit 13 (re-evaluation unit). Details of the revision suggestion information and the display contents on the user terminal device 2 will be described later.

<Processing Procedure>
Next, referring to FIG. 4, the details of the process from creating a draft of a contract to proposing a revision in the document revision system of this embodiment will be described. FIG. 4 is a sequence diagram showing an example of a processing procedure by the document revision device 1, the user terminal device 2, and the text generation device in the document revision system of this embodiment. Note that the process shown here is an example, and omissions and modifications may be made without changing the gist of the present invention. In addition to contracts, documents that may be targeted include other legal documents such as regulations, and other documents. In the following example, both the creation of a draft and the creation of a revision proposal described below are performed using the same text generation model, but these may be created using different models.

(1) Draft Creation Steps S401 to S406 are processes related to the creation of a draft. Note that this process is not essential, and the processes from step S407 onward may be performed using a draft created by the user himself/herself, a draft generated using an external machine learning model, or a draft contract received from a party to whom a contract is to be concluded.

First, in step S401, the user terminal device 2 accepts input of advance information from the user and transmits it to the document correction device 1. Assumed advance information includes the relationship with the other party to whom the contract is to be concluded, the purpose of the contract, the type of the contract, and the requests of the parties regarding the contract. Specifically, when a confidentiality agreement is concluded when presenting information to a business partner, examples of the relationship with the other party include "discloser of information," the type of contract includes the contract type "confidentiality agreement," and the requests of the parties include arbitrary requests such as "I want to make a strong contract (favorable to my position)." It is also preferable that the names or titles of the contracting parties and the designations in the contract (Party A, Party B) are also input as advance information. As advance information, term sheet information indicating the contents of a term sheet used in contract negotiations, questions about the contents to be included in the legal document, and question information based on the answers to those questions may be further acquired.

These advance information are typically entered by the user selecting from a selection of options. Alternatively, the information may be entered by the user freely writing text in a natural language.

Next, in step S402, the command generation unit 12 generates a creation command for causing the sentence generation model to create a draft based on the prior information. Since the sentence generation model of this embodiment is a conversational model that receives a sentence in natural language as input and generates a sentence as a response, the creation command is generated as a sentence in natural language.

More specifically, the command generating unit 12 can generate a creation command by generating a sentence that combines a prepared instruction sentence and a wording specified based on the prior information. In addition, for example, a configuration may be adopted in which an instruction sentence is prepared in advance for each option of the prior information and stored in a storage unit of the document correction device 1, an instruction sentence is selected from the prior information according to the contract type, and the command generating unit 12 generates a sentence that combines the instruction sentence and the wording specified based on the prior information as a creation command.

More specifically, when the prior information is input by selection, first, the word corresponding to the selected prior information is identified. For example, by setting a corresponding word in advance for each option of prior information, the word corresponding to the selected prior information can be identified. Then, the command generation unit 12 generates a creation command based on those words and an instruction statement separately registered in advance in the storage unit of the document correction device 1. On the other hand, when the prior information is input as free text, the creation command can be generated using the input prior information and an instruction statement prepared in advance.

Here, it is preferable to store a character string describing a document creation rule according to the prior information as a word corresponding to the prior information. For example, in a legal document, the legal meaning may differ between describing an exception in a sentence of a provision and describing a disclaimer separately after the sentence. By combining words that specify rules for such description methods according to the prior information, an appropriate legal document can be created. Note that, of the multiple document creation rules corresponding to each piece of prior information, the document correction device 1 may determine the document creation rule to be included in the creation command based on the prior information, or the document creation rule to be used by the text generation model may be determined by including multiple document creation rules in the creation command and including a statement in the creation command that an appropriate document creation rule will be used based on the prior information from the multiple document creation rules.

For example, instructions such as "Please create a document based on the following conditions" or "Enter prior information indicating the contractual arrangements. Please create an appropriate contract document based on the following rules" can be stored in advance in the storage unit of the document correction device 1. Alternatively, instructions can be prepared for each contract type as a selection of prior information, and information corresponding to the prior information can be included in the instructions, such as "Please create a non-disclosure agreement based on the following conditions." Then, by generating a sentence in which the instructions are followed by words based on the prior information, a creation command can be generated to instruct the text generation model to create a draft.

Once the creation command has been generated, the process proceeds to step S403, where the command generation unit 12 transmits the generated creation command to the text generation device 3. The transmission of the creation command to the text generation device 3 in step S403 corresponds to the transmission of the creation command to the text generation model. Upon receiving the creation command, the text generation device 3 inputs the creation command into the text generation model. This causes the text generation model to output a draft of the contract (step S404).

The text generation device 3 transmits the draft output by the text generation model to the document correction device 1 in step S405. Then, in step S406, the document correction device 1 transmits the draft to the user terminal device 2. Through the above process, the user can obtain a draft of the desired contract by only inputting the advance information. Note that step S406 and steps S407 to S410 described below may be omitted, and the process may be carried out in a consistent manner up to the revision of the draft without waiting for instructions from the user.

(2) Draft Evaluation Next, the evaluation process of the draft will be described. Steps S407 to S409 are processes related to the evaluation of the draft. This process is for acquiring evaluation information indicating the evaluation of the draft. In this embodiment, as described below, the evaluation unit 13 evaluates the draft, but the evaluation information may be acquired by having an external device evaluate the draft, or the evaluation information may be acquired by receiving the evaluation information together with the draft information from the user terminal device 2.

In step S407, the user terminal device 2 transmits an evaluation request to the document correction device 1. The evaluation request includes draft information and background information related to the draft. The draft information transmitted here is information including at least the contents of the draft to be evaluated, and in this embodiment, a document file containing a draft of the contract is uploaded. The background information is information related to the background of the document. For example, the background information includes the relationship with the other party in the arrangement indicated by the target legal document, the type of arrangement, the requests of the parties regarding the arrangement, etc. The background information can be information of the same content and format as the advance information in draft creation described above.

In the document correction device 1, the acquisition unit 11 receives an evaluation request, and the evaluation unit 13 performs evaluation of the draft based on the draft information and background information. The evaluation of the draft may be performed by any method, such as evaluation using a machine learning model or rule-based evaluation. As an example, in this embodiment, a method of evaluating clauses based on rules in accordance with background information is used.

In particular, in this embodiment, the contract type and contractual position from the background information, and the title of each clause in the draft are used to evaluate the content of each clause and the relationships between multiple clauses. Here, multiple clauses refers to a group of multiple clauses that prescribes a set of regulations. For example, within a group, the presence or content of clause B is determined based on the content of clause A, and clause C is then confirmed based on that result. By specifying that the evaluation process branches depending on the content, it is possible to evaluate the relationships between multiple clauses on a rule-based basis.

The draft is evaluated based on whether it contains unnecessary clauses (or clauses that are disadvantageous to the party in terms of background information on the other party), whether there are clauses that should be added, whether the content of the clauses is appropriate, etc.

In this embodiment, the issues raised in the draft are identified by linking the issues with the above rules and registering them in advance in the storage unit of the document correction device 1. Specifically, for each issue, the above-mentioned evaluation rule, the type of issue (either deletion, addition, or content modification of a clause), the details of the issue, and the risk level are stored in advance in the storage unit.

The comments are statements proposing revisions or pointing out problems with the draft contents, such as "Is it necessary to stipulate a consultation clause?" or "Isn't the period for which confidentiality obligations are imposed set longer than necessary?" In addition, the comments may include explanatory text about the clause related to the comment or template information for that clause. The risk level is information that indicates the degree of risk assumed for each comment. The risk level may be fixed (unchangeable) on the system side, or may be changeable by user input.

Examples of risk level settings are as follows. Issues that are likely to become problems after the contract is concluded and that are related to important rights and obligations of the company can be set to risk level "A". Examples include "intellectual property ownership", "scope of damages", and "definition of confidential information". Issues that are unlikely to become problems even if they are missing can be set to risk level "C". Examples include "confidential information exceptions" and "intermediate reporting obligations". Issues that do not fall into A and C among multiple groups can be set to risk level "B". Note that risk levels may be set for each issue, or may be risks determined based on the wording of the clause. For example, risk levels may be set based on a judgment of whether the company is favorable or unfavorable based on the wording of the clause.

As described above, the evaluation unit 13 executes the evaluation process and creates evaluation information indicating the evaluation of the draft. The evaluation information includes, for each of one or more issues identified in the above process, the clause in the draft that corresponds to the issue, the content of the issue, and a risk level. When evaluation information is obtained from an external device or the user terminal device 2, evaluation information is also obtained for each issue, including the clause in the draft that corresponds to the issue, the content of the issue, and a risk level. As described above, the content of the issue includes information written in natural language, and this description corresponds to the evaluation statement in the present invention.

The contract may be evaluated using a text generation model. The text generation model may be the same as or different from the text generation model used for draft creation or revision. For example, the document revision device 1 acquires draft information and background information, reads out evaluation rules to be used for contract evaluation that are stored in advance based on the background information, generates evaluation commands based on the draft information and evaluation rules according to the background information, transmits the evaluation commands to the text generation device 3, and acquires the evaluation results (evaluation information) of the contract generated by the text generation device 3. The evaluation commands are generated as text in natural language. This is expected to reduce the effect of reducing the impact on evaluation due to variations in the writing of text and differences in expression. In addition, it is expected that an appropriate evaluation can be performed by using evaluation rules (e.g. evaluation rules set for each contract type) read out based on background information (e.g. contract type) from among multiple evaluation rules.

Furthermore, evaluation using the sentence generation model and evaluation using the rule base may be performed simultaneously. In this case, if the evaluation using the sentence generation model and the evaluation using the rule base differ, it is preferable to perform processing that prioritizes the evaluation using the rule base. This makes it possible to guarantee a minimum evaluation using the rule base even if there is uncertainty in the evaluation using machine learning. Note that the document correction device 1 may determine which evaluation rule to include in the correction command based on the background information from among multiple evaluation rules corresponding to each piece of background information, or may be configured to determine the evaluation rule to be used by the sentence generation model by including multiple evaluation rules in the evaluation command and including a statement to use an appropriate rule from the multiple rules based on the background information in the evaluation command.

When the evaluation unit 13 finishes evaluating the draft, in step S409, the evaluation unit 13 transmits evaluation information to the user terminal device 2. When the user terminal device 2 receives the evaluation information, the evaluation information is processed for display, and the processing result is displayed on the display.

Figure 5 shows an example of an evaluation screen displayed on the user terminal device 2 as a result of the evaluation information display process. The evaluation screen W5 displays an overview of the document draft and an overview of the evaluation results at the top of the screen. The contents of the draft identified by the draft information are displayed on the left side of the screen. In addition, a "Edit read contents" button is provided, and if there is an error in the contents identified from the draft information, the contents can be edited by selecting this button.

The right side of the screen displays the evaluation results based on the evaluation information. For each issue, the type of issue (whether it is a clause deletion, addition, or content modification), an explanation, and a template for the applicable clause are displayed.

For each issue, a differently shaped indication mark W51 is attached for each type of indication, and a serial number for each type is displayed. Similar indication marks W51 are also displayed along with the serial number in the part of the draft on the left side of the screen that corresponds to the indication, allowing the user to recognize the position in the draft that corresponds to the indication, and the corresponding part in the draft is highlighted or otherwise emphasized. In the example in Figure 5, circles indicate indications that the content of clauses should be revised, and triangles indicate indications that clauses are missing.

The indication mark W51 is displayed in a different color depending on the risk level. By operating the risk filter operation section W52, the indications to be displayed can be filtered according to the risk level. In FIG. 5, the state is displayed in which indications with a risk level of "B" or higher, that is, indications of A or B, are displayed.

A batch correction button W53 and an individual correction button W54 are also displayed together with the evaluation results. When the user selects one of these buttons, a correction request is sent in step S410 in FIG. 4. The scope of the correction instructions varies depending on the button selected.

There are three types of batch correction buttons W53: "Create revision proposals for all items at once," "Create revision proposal for item A," and "Create revision proposals for items A and B." When "Create revision proposals for all items at once" is selected, a correction request is sent for all issues. When "Create revision proposal for item A" is selected, a correction request is sent for issues with a risk level of A. When "Create revision proposal for items A and B" is selected, a correction request is sent for issues with a risk level of A and issues with a risk level of B. When the individual correction button W54 displayed for each issue is selected, a correction request is sent for that issue.

In addition, in this embodiment, when the Bulk Correction button W53 or the Individual Correction button W54 is selected, further requests for corrections from the parties are accepted. For example, for issues where multiple correction policies are expected, the user can specifically select one of them to accept a request for correction. Alternatively, when requesting corrections for multiple issues using the Bulk Correction button W53, a configuration may be used in which the user is prompted to select a rough policy, such as "bullish (highly favorable to the company)," "neutral," or "bearish (more favorable to the company than before the correction, but more favorable to other companies)." At this time, a correction command may be generated that generates a correction proposal for only the policy selected by the user, or a correction command may be generated that generates a correction proposal for all policies. Such requests may also be entered as free text.

(3) Correction Process Next, the correction process of the draft will be described. When a correction request is received as described above, the acquisition unit 11 acquires the draft information and the evaluation information in step S410. The acquisition unit 11 of this embodiment further acquires the above-mentioned background information and information on the request for correction sent together with the correction request in step S410. In this embodiment, the draft received in step S405, the background information received in step S407, and the evaluation information generated by the evaluation unit 13 in step S408 are stored in the storage unit, and the acquisition unit 11 acquires the draft information, the evaluation information, and the background information based on them. Note that when the draft is created and evaluated by an external device, the acquisition unit 11 may acquire the draft information and the evaluation information from the user terminal device 2 or the external device.

When a revision request specifies that only part of the draft should be revised, rather than the entire draft, the draft information does not need to include the contents of the entire draft, and only the parts related to the subject of the revision may be included in the draft information. However, for example, when there is a group of multiple clauses that define a set of regulations, and revisions are to be made to the issues raised in the clauses included in that group, it is preferable that the draft information includes the contents of the entire multiple clauses in that group, rather than just the single clause that corresponds to the issues raised.

Next, in step S411, the command generating unit 12 generates a revision command for causing the sentence generation model to generate a proposed revision of the draft based on the draft information, evaluation information, background information, and information on the parties' requests regarding revisions acquired by the acquiring unit 11. Since the sentence generation model of this embodiment is a conversational model that receives input sentences in natural language and generates answer sentences, the revision command is generated as a sentence in natural language.

The command generation unit 12 of this embodiment generates a sentence as a correction command that describes instructions regarding correction for each clause that corresponds to the pointed out item. In particular, when there are multiple pointed out items in a single clause or a group of clauses that defines a set of regulations, if a correction command is generated for each pointed out item, there is a possibility that multiple conflicting correction proposals will be generated. Therefore, the command generation unit 12 of this embodiment generates a sentence of a correction command for each clause or group of clauses that corresponds to the multiple pointed out items as a correction command for a clause or group of clauses that has multiple pointed out items.

The generation of correction commands will be specifically described below. First, regarding background information, when background information is input by selection, the words corresponding to the selected background information are first identified. For example, by presetting corresponding words for each background information option, it is possible to identify the words corresponding to the selected background information.

Similarly, when request information is input by selection, the words corresponding to the selected request are first identified. For example, by presetting corresponding words for each combination of the target issue and request options, it is possible to identify the words corresponding to the selected request for each issue.

Then, the command generation unit 12 generates a correction command based on those words, the draft information, and a command statement that has been separately registered in advance in the storage unit of the document correction device 1. On the other hand, when the background information and the request are input as free text, the correction command can be generated using the input background information and a command statement prepared in advance.

Here, as in the case of generating a creation command in draft creation, it is preferable to store a character string describing a document creation rule corresponding to the background information as a word corresponding to the background information. For example, in a legal document, the legal meaning may differ between describing an exception in a sentence of a provision and describing a disclaimer separately after the sentence. By combining words that specify rules for such description methods according to the background information, an appropriate legal document can be created. Note that, of the multiple rules corresponding to each piece of background information, the document correction device 1 may determine which rule to include in the correction command based on the background information, or the document generation model may determine which rule to use by including multiple rules in the correction command and including in the correction command a statement to use an appropriate rule from the multiple rules based on the background information.

As an instruction sentence, for example, "As a result of reviewing the above draft, the following points were made. Please create a revised version based on the following conditions, taking into account the points made," or "As a result of reviewing the above draft, the following points were made. Please create an appropriate contract document based on the following rules, taking into account the points made," etc., can be stored in advance in the storage unit of the document correction device 1. Then, a sentence that combines the contents of the draft, the instruction sentence, and words based on the background information and the request can be generated as a creation command to instruct the text generation model to create a draft.

Once the revision command has been generated, the process proceeds to step S412, where the command generation unit 12 transmits the generated revision command to the text generation device 3. The transmission of the revision command to the text generation device 3 in step S412 corresponds to the transmission of the revision command to the text generation model. The revision commands generated as described above for each clause or for each group of clauses may each be transmitted in multiple batches. For example, the configuration may be such that the directive portion of the revision command is transmitted first, followed by words corresponding to the issues raised or requests. Alternatively, the configuration may be such that it is determined whether to transmit the revision command all at once or in multiple batches depending on the number of characters in the revision command.

When the writing generation device 3 receives the revision command, it inputs the revision command to the writing generation model. As a result, the writing generation model outputs a revision proposal for the draft (step S413). In this embodiment, the evaluation unit 13 (re-evaluation unit) obtains a revision proposal for each pointed out item. Note that the instruction generation unit 12 may generate a revision command instructing the unit to propose multiple revision proposals for one pointed out item in step S411, and the evaluation unit 13 (re-evaluation unit) may obtain multiple revision proposals for some or all of the pointed out items in step S413.

Furthermore, a revision command may be sent to the text generation model to generate revision reasons based on the evaluation information, and the revision reasons may be obtained in addition to the revision proposal. Furthermore, revision reasons for one's own company (or for one's own client) and revision reasons for the other party may be generated and obtained. In contract negotiations, revisions are made repeatedly between one's own company and the other party, and at least one of revision reasons for explanation within the company and revision reasons for explanation outside the company is often written. Writing revision reasons requires a high level of expertise, but the above configuration can help even users without specialized knowledge to write revision reasons.

(4) Re-evaluation process In step S414, the writing generation device 3 transmits the proposed revision, which is the output of the writing generation model, to the document correction device 1. In the document correction device 1, re-evaluation is performed by the evaluation unit 13 (re-evaluation unit) in step S415. The evaluation unit 13 (re-evaluation unit) acquires the proposed revision, and re-evaluates the proposed revision in the same procedure as the evaluation process described above. Then, re-evaluation information indicating the evaluation of the proposed revision is generated in the same manner as the evaluation information described above. In other words, the re-evaluation information is information indicating the evaluation of the proposed revision, and the information included in the re-evaluation information is the same items as the evaluation information described above.

In this case, the reevaluation may be limited to only those parts of the draft that are related to the issues raised earlier, rather than the entire draft. Specifically, the clause or group of clauses indicated in the proposed amendments may be reevaluated, and the other parts may be omitted. This is expected to have the effect of improving processing speed.

Then, the evaluation unit 13 (reevaluation unit) judges whether the evaluation satisfies a preset criterion based on the reevaluation information. Specifically, it checks the risk levels of the items pointed out in the reevaluation information, and judges that the criteria are met if the risk levels of all the items pointed out are equal to or lower than a predetermined risk level.

If the reevaluation of the revision proposal does not meet the criteria, i.e., if there are any issues with a risk level that exceeds the specified risk level, the process returns to step S411 and the process from generating the revision command to reevaluating in step S415 is performed again. In this case, in the second and subsequent revision command generation processes, revision proposal information is used instead of draft information, and reevaluation information is used instead of evaluation information, and revision commands related to the revision proposal are generated in the same manner as the generation of the revision commands described above.

In principle, the loop process is repeated until a preset criterion is met, but if the number of repetitions becomes extremely high or it becomes impossible to exit the loop, there is a concern that the processing load will increase. Therefore, in this embodiment, an upper limit to the number of times the loop process is repeated is set as the termination condition, and even if the reevaluation does not meet the criterion, if the termination condition is met, the repeated reevaluation ends, the loop process ends, and the process proceeds to step S417. Other termination conditions may also be set, such as, for example, comparing the drafts before and after revision and the number of changed characters being below a certain level. This makes it possible to prevent a situation in which only minor revisions are made from occurring repeatedly.

On the other hand, if the evaluation of the proposed revision meets the preset criteria, the sequence diagram in FIG. 4 enters a loop break process (break process). In step S416, the suggestion unit 14 generates revision proposal information and ends the loop process (Loop process).

In the interruption process, the suggestion unit 14 generates revision proposal information. For each issue raised in the reevaluation, the revision proposal information includes the clause in the draft that corresponds to the issue, the content of the issue, the risk level, and the revision proposal created in the most recent step S413.

Then, in step S417, the suggestion unit 14 transmits the revision suggestion information generated in step S416 to the user terminal device 2. The user terminal device 2 displays the display processing result of the revision suggestion information on the display. This allows the user to receive suggestions for revising the draft.

The above reevaluation process procedure will be further explained with reference to the flowchart in FIG. 6, particularly with regard to the conditions for repetition. When the document correction device 1 receives the proposed revision, which is the output of the sentence generation model, in step S414 in FIG. 4, the evaluation unit 13 (reevaluation unit) acquires this proposed revision in step S501.

Then, in step S502, a reevaluation is performed (corresponding to step S415 in FIG. 4), and in step S503, it is determined whether the evaluation of the revision proposal satisfies a preset criterion. As described above, in this embodiment, the risk levels of the identified items in the reevaluation information are checked, and it is determined that the criterion is met if the risk levels of all identified items are equal to or lower than a predetermined risk level.

If the criteria are not met, the process proceeds to step S505, where it is confirmed whether the number of loops (number of revisions) has reached the upper limit. If the upper limit has not been reached ("Y" in step S505), the process proceeds to step S506, where the evaluation unit 13 (re-evaluation unit) causes the command generation unit 12 to generate and transmit a revision command again, increments the loop count counter, and the process returns to step S501.

If the criteria are met in step S503, or if the number of revisions reaches the upper limit in step S505, the process proceeds to step S504, where the evaluation unit 13 (re-evaluation unit) causes the proposal unit 14 to generate revision proposal information (corresponding to step S416 in FIG. 4). The proposal unit 14 then transmits the revision proposal information to the user terminal device 2 (corresponding to step S417 in FIG. 4), and the process ends. Note that the above processing procedure is an example, and may be changed as desired, for example by switching the judgment order of steps S503 and S505.

(5) Revision Proposal Screen Next, a revision proposal screen displayed on the user terminal device 2 as a result of the display process of the revision proposal information will be described with reference to Fig. 7. Fig. 7 is a display example of the revision proposal screen in this embodiment. The revision proposal screen W7 in this embodiment is displayed in a similar configuration to the evaluation screen W5 shown above, and the indication mark W71 and the risk filter operation section W72 are similar to the indication mark W51 and the risk filter operation section W52 in the evaluation screen W5, and therefore description thereof will be omitted.

The difference from the evaluation screen W5 is mainly in the content displayed on the right side of the screen, particularly the section related to "revision proposals." The revision proposal screen W7 includes revision proposals W73 for each pointed out item as a display based on the revision proposals included in the revision proposal information. The revision proposals W73 display the above-mentioned revision proposals and also display the text so that it can be copied. It is also preferable to make it easy for the user to recognize by highlighting the changes from the draft. This allows the user to obtain the content and revision proposals for each pointed out item, and even users who are not familiar with legal knowledge can make revisions to legal documents on their own.

When multiple revision suggestions are obtained for a problem, the revision suggestion information contains multiple revision suggestions for the problem, so it is preferable to display multiple revision suggestions W73 for the problem on the revision suggestion screen W7 as well, so that the user can select from them.

The method of outputting the revision proposal information is not limited to displaying the revision proposal screen W7 described above. For example, the suggestion unit 14 may transmit to the user terminal device 2 a document-format file of the entire draft to which revision proposals have been applied to at least some of the pointed out items. In addition, the user may be prompted to select whether or not to reflect the revision proposals displayed on the revision proposal screen W7, or which revision proposals to reflect, and the suggestion unit 14 may further transmit to the user terminal device 2 a document-format file in which the entire draft has been revised based on the result of the selection.

As described above, the document revision system of this embodiment can assist users who do not have specialized knowledge in making revisions to draft legal documents such as contracts. Furthermore, the quality of the draft revisions can be improved by reevaluating the revision proposal and issuing a command to create another revision proposal based on the results.

In addition, by generating commands in natural language text, it is possible to create revision proposals using a conversational text generation model that takes text as input and generates answer text. This makes it possible to have the model generate revision proposals that take into account various conditions such as the background and requests of the contract, making it possible to provide draft revision proposals that are more in line with the user's situation and wishes.

Furthermore, a revision command may be sent to the text generation model to generate revision reasons based on the evaluation information, and the revision reasons may be included in the revision proposal information. Furthermore, revision reasons for one's own company (or for one's own client) and revision reasons for the other party may be generated. In contract negotiations, revisions are made repeatedly between one's own company and the other party, and at this time, at least one of revision reasons for explanation within the company and revision reasons for explanation outside the company is often written. Writing revision reasons requires a high level of expertise, but the above configuration can help even users without specialized knowledge to write revision reasons.

0: document correction system 1: document correction device 2: user terminal device 3: text generation device 11: acquisition unit 12: command generation unit 13: evaluation unit 14: proposal unit 90: terminal device 101: control unit 102: memory unit 103: communication unit 901: control unit 902: memory unit 903: communication unit 904: input unit 905: output unit NW: communication network W5: evaluation screen W51: indication mark W52: risk filter operation unit W53: collective correction button W54: individual correction button W7: correction proposal screen W71: indication mark W72: risk filter operation unit W73: correction proposal

Claims (20)

1. A document revision program for utilizing a sentence generation model to suggest revisions to a legal document based on an evaluation of the legal document, the program comprising:
A computer is configured to function as an acquisition unit, an evaluation unit, an instruction generation unit, and a suggestion unit;
The evaluation unit generates evaluation information including a plurality of legal findings for the draft legal document;
The acquisition unit acquires draft information indicating the content of a portion of the description in the draft that corresponds to a specified issue selected by a user from among the multiple specified issues , and a portion of the evaluation information that corresponds to the specified issue ,
the instruction generation unit generates a revision instruction for the draft based on a part of the draft information and the evaluation information acquired by the acquisition unit , and transmits the revision instruction to the writing generation model;
The suggestion unit transmits correction suggestion information for correcting a part of the draft corresponding to the pointed out item selected by the user based on the output of the sentence generation model. The instrument includes a plurality of clauses;
The document correction program of claim 1 , wherein the evaluation information includes a result of evaluating the draft with respect to each clause or a relationship between a plurality of clauses. The document correction program according to claim 2, wherein the evaluation information includes an evaluation of any of unnecessary clauses, missing information, and clause contents. The document correction program according to claim 2 or 3, wherein the instruction generation unit generates, as the correction instruction, a sentence in which instructions regarding the correction are written for each clause or for each group of multiple clauses that defines a set of regulations, and transmits the sentence to the sentence generation model. The evaluation information includes a risk level for each of the issues pointed out in the draft,
The document correction program according to claim 1 , wherein the instruction generating unit generates, as the correction instruction, an instruction to correct the draft for an issue that is at or above a predetermined risk level, and transmits the instruction to the text generation model. the evaluation information includes an evaluation statement describing an evaluation of the draft in a natural language;
The document correction program according to claim 1 , wherein the instruction generating unit generates a sentence based on the draft, the evaluation sentence, and a prepared instruction sentence as the correction instruction and transmits the correction instruction to the sentence generation model. The computer further functions as a re-evaluation unit;
The re-evaluation unit is
obtaining a proposed revision of the draft as an output of the sentence generation model;
obtaining reassessment information indicative of an evaluation of the proposed revision and determining whether the evaluation satisfies a predefined criterion;
If the criteria are met, the suggestion unit transmits the modification suggestion information;
2. The document correction program according to claim 1, further comprising: a command generating unit configured to generate a correction command for the proposed revision based on correction information and reevaluation information for the proposed revision and transmit the generated correction command to the sentence generation model if the criterion is not satisfied. The document correction program according to claim 7, wherein the reevaluation unit causes the suggestion unit and the instruction generation unit to repeatedly obtain reevaluation information and send correction instructions until the evaluation of the correction proposal satisfies a preset criterion. The reassessment information includes a risk level for each of the issues identified with respect to the proposed revision;
9. The document correction program of claim 8, wherein the criteria includes a risk level of all issues for the proposed correction being at or below a predetermined risk level. The document correction program according to claim 8 or 9, wherein the re-evaluation unit, when a preset termination condition is met, terminates the re-evaluation repetition and causes the suggestion unit to transmit the correction proposal information even if the evaluation of the correction proposal does not satisfy a criterion. The document correction program according to claim 1, wherein the command generation unit receives input of a correction request from a user, generates the correction command based on the request, and transmits the correction command to the text generation model. The legal document is a legal document relating to an agreement with a specific party;
The document correction program according to claim 1 , wherein the instruction generation unit acquires background information indicating at least one of a relationship with the other party, a type of the agreement, and requests of the parties regarding the agreement, generates the correction instruction based on the background information, and transmits the correction instruction to the text generation model. The document correction program of claim 1 , wherein the draft is a model-generated draft . The legal document is a legal document relating to an agreement with a specific party;
14. The document correction program of claim 13, wherein the draft is a draft generated by the model based on prior information indicative of at least one of a relationship between the parties, a type of the agreement, and a desire of the parties regarding the agreement. The computer further functions as a user terminal device ;
the evaluation unit executes an evaluation of the draft , generates information including one or more pointed out items regarding the draft as the evaluation information, and transmits the generated evaluation information to the user terminal device, thereby causing the user terminal device to display an evaluation screen including the pointed out items;
A document correction program as described in any one of claims 1 to 3, wherein the acquisition unit acquires the draft information corresponding to the issue selected on the evaluation screen displayed on the user terminal device and a portion of the evaluation information corresponding to the issue. the evaluation screen has a correction button selectable by a user for each of the indicated items,
The document correction program according to claim 15, wherein the instruction generation unit generates a correction instruction for the draft based on the draft information and evaluation information related to the indication corresponding to the correction button selected by the user, and transmits the correction instruction to the text generation model. The document correction program according to claim 16, wherein the suggestion unit causes the user terminal device to display an evaluation screen that displays, based on the output of the sentence generation model, a suggestion corresponding to the correction button selected by the user and a sentence indicating a correction suggestion based on the suggestion, in association with each other. The computer further functions as a user terminal device ;
2. The document correction program according to claim 1, wherein the suggestion unit transmits the correction proposal information including a plurality of correction proposals for at least one of the pointed out items to the user terminal device, and causes the user terminal device to display an evaluation screen including the plurality of correction proposals for at least one of the pointed out items. 1. A document correction device for proposing corrections to a legal document based on an evaluation of the legal document by utilizing a sentence generation model, comprising:
The system includes an acquisition unit, an evaluation unit, an instruction generation unit, and a suggestion unit,
The evaluation unit generates evaluation information including a plurality of legal findings for the draft legal document;
The acquisition unit acquires draft information indicating the content of a portion of the description in the draft that corresponds to a specified issue selected by a user from among the multiple specified issues , and a portion of the evaluation information that corresponds to the specified issue ,
the instruction generation unit generates a revision instruction for the draft based on a part of the draft information and the evaluation information acquired by the acquisition unit , and transmits the revision instruction to the writing generation model;
The suggestion unit transmits correction suggestion information for correcting a part of the draft corresponding to the pointed out item selected by the user based on the output of the sentence generation model. 1. A method of document correction utilizing a sentence generation model to suggest corrections to a legal document based on an evaluation of the legal document, comprising:
The computer
generating evaluation information including a plurality of legal findings for the draft legal document;
Obtaining draft information indicating the content of a portion of the description in the draft that corresponds to a problem selected by a user from among the plurality of problems , and a portion of the evaluation information that corresponds to the problem ;
generating revision instructions for the draft based on a portion of the draft information and the evaluation information and sending the revision instructions to the writing generation model;
A document correction method, comprising: transmitting correction suggestion information for correcting a part of the draft corresponding to the pointed out matter selected by the user based on an output of the sentence generation model.