Support program, support system, and support method

Abstract

To solve a conventional problem in which a document is corrected by partial replacement to a recommended document registered in advance, which is unable to deal with only a content anticipated in advance and is difficult to flexibly carry out correction tailored to each document.SOLUTION: A support program for supporting creation or correction of a target document which is a legal document to be created or corrected, causes a computer to function as an acquisition unit, a generation unit, and a proposal unit. The acquisition unit acquires item information regarding description contents corresponding to insufficient items that are insufficient in the target document, and supplementary information as information about the target document and for making a proposal corresponding to the target document. The generation unit inputs the item information and the supplementary information to a sentence generative model. The proposal unit outputs proposal information for supporting creation of the insufficient items in the target document, based on an output of the sentence generative model.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a support program, a support system, and a support method for assisting in the creation or modification of legal documents.

Background Art

[0002] In the evaluation and creation of documents, particularly legal documents such as contracts and regulations, high expertise is required. Against this background, conventionally, techniques for checking legal documents and techniques for assisting in the modification of legal documents based on such checks are known.

[0003] For example, Patent Document 1 describes a technique for receiving input of information regarding review results of legal documents and performing modifications. In the technique described in Patent Document 1, modifications to legal documents can be made by replacing corresponding parts with pre-registered recommended sentences.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, in the technique described in Patent Document 1, since modifications are made by partially replacing with pre-registered recommended sentences, it can only handle pre-assumed contents and it is difficult to flexibly make modifications according to individual documents. Therefore, the present invention makes the above problems the problems to be solved.

Means for Solving the Problems

[0006] To solve the above problems, the present invention provides a support program for assisting in creating or modifying a target document, where the target document is a legal document to be created or modified, and causes a computer to function as an acquisition unit, a generation unit, and a proposal unit. The acquisition unit acquires item information regarding the description content corresponding to the lacking items that are lacking in the target document, and information regarding the target document, which is auxiliary information for making a proposal according to the target document. The generation unit inputs the item information and the auxiliary information into a text generation model, and the proposal unit outputs proposal information for assisting in creating the lacking items in the target document based on the output of the text generation model.

[0007] With such a configuration, it is possible to assist in creating the uncreated parts in the target document. Also, by inputting the description examples and the auxiliary information into the text generation model and outputting the proposal information based on the output thereof, it is possible to assist in creating a document according to the target document, rather than just presenting a template.

[0008] In a preferred form of the present invention, the description examples include description examples of legal documents corresponding to the lacking items, and the description examples are descriptions or templates in documents that define past agreements.

[0009] With such a configuration, it is possible to prepare in advance the descriptions corresponding to the lacking items and make proposals regarding the creation and modification of the target document based on this.

[0010] In a preferred form of the present invention, the computer is further caused to function as an evaluation unit. The evaluation unit acquires evaluation information indicating the evaluation result of the target document, and the acquisition unit acquires the description examples based on the evaluation information.

[0011] With such a configuration, it is possible to make a proposal using an appropriate description example according to the evaluation result of the target document.

[0012] In a preferred embodiment of the present invention, the evaluation information includes an item type indicating the type of the lacking item, and the acquisition unit acquires the description example corresponding to the item type of the lacking item from a database that stores description examples indicating the description content associated with the item type.

[0013] With such a configuration, based on the description example according to the item type, it is possible to make proposals regarding the creation and modification of the target document.

[0014] In a preferred embodiment of the present invention, the evaluation information includes an evaluation comment regarding the lacking item, and the acquisition unit acquires the evaluation comment as the item information.

[0015] With such a configuration, for example, based on the evaluation comment describing the correction policy or the like, it is possible to make proposals regarding the creation and modification of the target document.

[0016] In a preferred embodiment of the present invention, the evaluation information includes one or more pointed-out matters regarding the target document, and a pointed-out type selected from a plurality of groups including item deficiency and clause modification is assigned to the pointed-out matters. The acquisition unit determines the pointed-out type of the pointed-out matters included in the evaluation information, and for the pointed-out matters of item deficiency, acquires the item information of the lacking item related to the pointed-out matters, and for the pointed-out matters of clause modification, acquires target document information indicating the description content of the target document related to the pointed-out matters. The generation unit inputs the item information and auxiliary information into the text generation model for the pointed-out matters of item deficiency, while inputs the target document information and information regarding the agreement defined by the target document into the text generation model for the pointed-out matters of clause modification.

[0017] With such a configuration, appropriate proposals can be made according to the pointed-out type. Specifically, for item deficiency, it is possible to support the creation of the lacking item based on the description example and auxiliary information. On the other hand, for clause modification, it is possible to support the modification of the part related to the pointed-out matters based on the target document information indicating the description content of the target document.

[0018] In a preferred embodiment of the present invention, the target document includes a plurality of clauses, and the types of pointed-out deficiencies indicating a lack of items are divided into a plurality of types including a partial deficiency in which some of the clauses within a group defining a set of regulations are lacking, and a complete deficiency in which the entire group of clauses is lacking. The generation unit determines the input to the text generation model according to whether the pointed-out matters related to a group of a plurality of clauses are a partial deficiency or a complete deficiency.

[0019] With such a configuration, even when defining a set of regulations with a plurality of clauses, an appropriate proposal can be made.

[0020] In a preferred embodiment of the present invention, when the type of pointed-out deficiency related to the lack of items is the partial deficiency, the acquisition unit acquires, as the auxiliary information, the description content of other clauses belonging to the same group as the lacking item in the target document.

[0021] With such a configuration, it becomes possible to make a proposal considering the relationship between a plurality of clauses.

[0022] In a preferred embodiment of the present invention, the auxiliary information includes target document information indicating at least a part of the content of the target document.

[0023] With such a configuration, based on the content of the target document, it is possible to make a proposal for creating or modifying the lacking item.

[0024] In a preferred embodiment of the present invention, the target document information includes information on the insertion position of the lacking item in the target document and the description content before and after the insertion position.

[0025] With such a configuration, based on the description content before and after the insertion position in the target document, an appropriate proposal can be made.

[0026] In a preferred form of the present invention, the target document includes a plurality of clauses, and the target document information includes the description content of the clauses in the target document related to the missing items.

[0027] With such a configuration, appropriate proposals can be made based on the description content of the clauses of the target document related to the missing items.

[0028] In a preferred form of the present invention, the target document is a legal document including a preamble, and the target document information includes the description content of the preamble in the target document.

[0029] In a legal document, the preamble may describe the premises such as the definitions of the words used in the document. With such a configuration, the content of the target document can be appropriately reflected in the proposal.

[0030] In a preferred form of the present invention, the auxiliary information includes definition information indicating the definitions of the words used in the target document.

[0031] With such a configuration, the content of the target document can be appropriately reflected in the proposal.

[0032] In a preferred form of the present invention, the auxiliary information includes style information indicating the style of the target document.

[0033] With such a configuration, proposals can be made with the content that unifies the style within the document.

[0034] In a preferred form of the present invention, the acquisition unit further acquires the correction intensity selected by the user, and the generation unit further inputs the correction intensity into the text generation model.

[0035] With such a configuration, proposals reflecting the intention of the user can be made.

[0036] In a preferred form of the present invention, the computer is further caused to function as a specifying unit, the acquisition unit further acquires an item type indicating the type of the lacking item and a correction intensity selected by the user, and the specifying unit specifies, based on the correction intensity, an example description to be used for creating or correcting the target document from among the example descriptions corresponding to the item type, from a database storing a plurality of example descriptions associated with the item type, and the generation unit inputs the example description and the auxiliary information specified by the specifying unit into the sentence generation model.

[0037] With such a configuration, a proposal can be made using an example description according to the intention of the user.

[0038] In a preferred form of the present invention, the generation unit generates a generation instruction for instructing to generate a text related to the lacking item based on the item information and the auxiliary information as a sentence in natural language, and inputs the generation instruction into the sentence generation model.

[0039] With such a configuration, for example, a proposal can be made using a conversational sentence generation model that outputs a sentence in response to the input of a sentence.

[0040] In a preferred embodiment of the present invention, a computer is further functioned as an evaluation unit. The evaluation unit acquires evaluation information indicating an evaluation result of the target document. The evaluation information includes one or more pointed-out matters regarding the target document. The pointed-out matters are given a pointed-out type selected from a plurality of groups including item deficiency and clause amendment. The acquisition unit determines the pointed-out type of the pointed-out matters included in the evaluation information, and for the pointed-out matters of item deficiency, acquires item information of the deficient items related to the pointed-out matters, and for the pointed-out matters of clause amendment, acquires target document information indicating the described content of the target document related to the pointed-out matters. The generation unit inputs, into the text generation model, a generation command that instructs to generate a text regarding the deficient items based on the item information and the auxiliary information for the pointed-out matters of item deficiency, and inputs, into the text generation model, a generation command that instructs to generate an amendment related to the pointed-out matters based on the target document information and the auxiliary information for the pointed-out matters of clause amendment.

[0041] With such a configuration, by using a conversational text generation model, appropriate proposals can be made according to the pointed-out type.

[0042] The present invention is a support system for assisting in the creation or amendment of legal documents, comprising an acquisition unit, a generation unit, and a proposal unit. The acquisition unit acquires item information regarding the described content corresponding to the deficient items that are deficient in a target document, which is a legal document to be created or amended, and auxiliary information regarding the target document for making a proposal according to the target document. The generation unit inputs the described example and the auxiliary information into a text generation model. The proposal unit transmits proposal information for assisting in the creation of the deficient items in the target document based on the output of the text generation model.

[0043] The present invention is a support method for assisting in the creation or modification of legal documents, which is executed by a computer comprising an acquisition unit, a generation unit, and a proposal unit. The acquisition unit acquires item information regarding the description content corresponding to the lacking items that are lacking in a target document, which is the legal document to be created or modified, and auxiliary information regarding the target document for making a proposal according to the target document. The generation unit inputs the description examples and the auxiliary information into a text generation model, and the proposal unit transmits proposal information for assisting in the creation of the lacking items in the target document based on the output of the text generation model.

Advantages of the Invention

[0044] According to the present invention, it is possible to appropriately support the creation or modification of a target document, including the creation of the lacking parts.

Brief Description of the Drawings

[0045]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Modes for Carrying Out the Invention

[0046] Hereinafter, the support system of the present invention will be described with reference to the drawings. Preferred embodiments are shown in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein.

[0047] For example, in this embodiment, the configuration, operation, etc. of the support system will be described. However, the same effects can also be achieved by a support device, a support system, a method executed by the support device, a support program, etc. In this embodiment, the support system takes a so-called cloud computing mode in which the support program is executed on a server (support device) connected via a network to a client terminal (user terminal) in order to realize its functions on the client terminal.

[0048] The support program may be provided to a computer device as a computer-readable non-transitory storage medium in order to function any computer device as a support device, or may be provided so as to be downloadable from an external server. Further, the document correction program may be a program that is add-on (plug-in) in document editing software such as Microsoft Word (registered trademark).

[0049] In this embodiment, the "unit" may include, for example, a combination of hardware resources implemented by a circuit in a broad sense and information processing of software that can be specifically realized by these hardware resources. In this embodiment, the "information" may be represented, for example, by a physical value of a signal value representing voltage or current, the high or low of a signal value as a binary bit aggregate composed of 0 or 1, or a quantum superposition (so-called quantum bit), and communication and calculation can be executed on a circuit in a broad sense. The circuit in a broad sense is circuitry realized by appropriately combining a circuit, a processor, a memory, etc. For example, it is circuitry including any of 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.

[0050] <1. Definition> First, the definition of terms will be explained. The present invention is a technology for assisting in creating or modifying a document by using a text generation model. The text generation model refers to any model capable of outputting text. As the text generation model, a model generated by machine learning can be typically used. In this embodiment, the case of using a conversational text generation model that generates text as an answer to the input text by inputting the text will be described.

[0051] More specifically, a text generation model is a machine learning model for natural language processing and is a model that has the function of generating text based on the input information. For example, a text generation model is a machine learning model based on the Transformer model having a self-attention mechanism. The Transformer model is, for example, a GPT (Generative Pre-trained Transformer) model. By executing learning with a large-scale dataset, the text generation model can perform not only general natural language generation tasks but also highly specialized text generation tasks.

[0052] For example, as the text generation model, a model learned with a large-scale dataset using the GPT-3 or GPT-4 algorithm developed by OpenAI (for example, the ChatGPT model generated by combining learning with a large-scale dataset and reinforcement learning using a reward prediction model) can be applied. Also, in this embodiment, an example of the input to the model is text information, but the input may be image information or voice information.

[0053] In the present invention, the target document refers to the document that the user wants to create or modify. In this embodiment, legal documents such as contracts, rules, and regulations are particularly assumed as the target document. Some legal documents include units divided into articles, paragraphs, items, etc., and in this embodiment, each of such units is called an "article or paragraph". In this embodiment, the case where the legal document is a contract is illustrated. Examples of contracts include documents that exert contract effects such as confidentiality agreements, sales contracts, and contracts related to real estate leases. Furthermore, this embodiment is not limited to contracts, and for example, the legal document may be a rule, regulation, notice, etc. Examples of rules include legal norm documents such as employment rules. Also, examples of notices include policy declaration documents such as privacy policies. In addition, legal documents include documents having some relevance to laws such as demand letters and letters of attorney.

[0054] When evaluating legal documents, the evaluation is carried out from various perspectives. Examples of the evaluation perspectives include whether the necessary items exist in the target document, whether there are no adverse conditions, and whether there are no unnecessary items, etc. The "item" in the present invention is one of such evaluation perspectives and is a unit for confirming deficiencies in the content of the target document. The present invention inputs information into a text generation model for the deficiency items that are deficient items in the target document, and makes proposals based on the output. For example, clauses may be used as the items of the present invention, or the arguments stipulated in legal documents may be used as the items of the present invention.

[0055] In legal documents including contracts, there are matters that can be debated or negotiated currently or in the future regarding their content, necessity, and other matters from a legal perspective. Generally, these are called arguments. Specifically, the arguments in the present invention can include the contract period, the subject matter of the contract, and each specific type of exception defined as an exception to confidential information in a confidentiality contract, etc. Also, when a set of regulations is defined by a plurality of clauses, the set of regulations can be regarded as one argument. In this embodiment, a form in which an argument is taken as an "item", that is, a form in which deficiencies are confirmed for each argument, will be exemplified and described.

[0056] Here, the "deficiency item" in the present invention widely includes items that are not created or are insufficient regardless of the creation status of the target document. That is, the deficiency item is not limited to what is considered deficient in the existing target document, but also includes items that should be included in the document to be created in the future. Therefore, the present invention can be used not only in the document modification stage, but also, for example, at the stage before a specific document draft is created, to create a text corresponding to the item that is desired to be included in the target document.

[0057] In the present invention, the auxiliary information is information regarding the target document and is information for making a proposal according to the target document. For example, target document information indicating at least part of the content of the target document, background information regarding the background of the target document, style information indicating the style of the target document, definition information indicating the definition of words used in the target document, evaluation information indicating an evaluation regarding the target document, etc. are used as auxiliary information. Regarding the background information, style information, and definition information, they may be generated and acquired based on the target document information. Also, regarding the target document information, when using part of the content instead of the entire text of the target document, it is preferable to use information indicating part of the content that can be particularly used to correct the item information described later according to the target document.

[0058] As the background information, information regarding the agreements stipulated by the target document in particular can be preferably used. More specifically, for example, in the agreement indicated by the target document, information such as the counterparty of the agreement, the relationship with the counterparty, the purpose of the agreement, the desires of the parties related to the agreement, information indicating the content to be agreed upon, the type of agreement, the desires of the parties related to the agreement, etc. is included in the background information.

[0059] As the style information, for example, information indicating whether the style of the target document is the polite form (desu / masu form) or the plain form (da form) can be used. The style information may be acquired by determining whether it is the polite form or the plain form based on the target document information.

[0060] The definition information is, for example, information indicating whether the company is A or B, or for example, the definition of terms used in the target document. The definition information may be specified by user input or may be acquired based on the target document information. Specifically, for example, the information inside parentheses indicates a definition, and the description in a predetermined format in the previous text is regarded as a definition, etc. The definition information can be acquired based on rules.

[0061] Auxiliary information other than the target document information (background information, style information, definition information, etc.) may be, for example, term sheet information input by a term sheet used in contract negotiation. The term sheet is a tabular arrangement of the content agreed upon with the other party or the proposed content regarding the main items of the contract, and the term sheet information is information including the content described in the term sheet. Further, the auxiliary information may be question information based on the question content and question results generated by question software that asks questions about the content to be described in a legal document (for example, a contract).

[0062] The present invention makes a proposal regarding the creation or modification of a target document by inputting the above-described auxiliary information together with item information regarding the description content corresponding to the missing items into a text generation model.

[0063] <2. System Configuration> FIG. 1 is a block diagram showing the configuration of the support system of the present embodiment. As shown in FIG. 1, the support system 0 includes a support device 1 and a user terminal 2. The support device 1 and the user terminal 2 are configured to be communicable via a network NW. The support device 1 operates as a server, and the user terminal 2 operates as a client terminal. In the present embodiment, the network NW is an IP (Internet Protocol) network, but there are no restrictions on the type of communication protocol, the type of network, etc.

[0064] Further, the support device 1 is connected to a generation device 3 via the network NW. The generation device 3 is an external server device that provides a text generation model. Note that the support device 1 may store the text generation model and be configured to be able to make a proposal regarding the creation or modification of a target document without using an external device, or a configuration may be adopted in which part or all of the processing executed by the support device 1 in the present embodiment is executed by the user terminal 2.

[0065] As the support device 1, the generation device 3, and the database server, one or more information processing devices 10 (computer devices) such as general-purpose servers and personal computers can be used. In this embodiment, the support device 1 is an information processing device 10 installed with a computer program (support program) that executes the support method. However, it may be made to function as the support device 1 by installing the support program on the terminal device 90.

[0066] As the user terminal 2, a terminal device 90 (computer device) such as a personal computer, a smartphone, or a tablet terminal can be used. The user terminal 2 has an application (web browser application or dedicated application) for making a request to the support device 1 and receiving a response, and uses the processing result of the support program executed in the support device 1 via this application.

[0067] Based on the request sent from the application of the user terminal 2, the support device 1 sends data in a predetermined format as various display information to the application of the user terminal 2. The data in the predetermined format is, for example, data in JSON (JavaScript (registered trademark) Object Notation) format or XML format. The application of the user terminal 2 renders the display data based on the received data and displays various types of information based on the proposed information on the display of the user terminal 2.

[0068] For example, when executing display processing of various types of information on the web browser application of the user terminal 2, first, the web browser application sends an HTTP (Hypertext Transfer Protocol) request to the web server and receives an HTTP response including the file requested by the HTTP request. The received file is display information including amendments, indicating the content or instructions of the web page to be displayed, and is described using a markup language such as HTML (Hypertext Markup Language), for example. A script, which is a program executed on the web browser application, may be added to the received file. The web browser application renders (draws) the web page in the display area based on the received file. When a script is added to the file, the browser executes the script and displays the web page according to the execution result.

[0069] <3. Hardware Configuration> Figure 2(a) is a hardware configuration diagram of the information processing apparatus 10. As shown in Figure 2, the information processing apparatus 10 includes a control unit 101, a storage unit 102, and a communication unit 103, which are used to exert the functions of each part and each process.

[0070] The control unit 101 has a processor such as a CPU capable of executing an instruction set, and executes an OS and programs. The storage unit 102 has a volatile memory such as a RAM capable of storing an instruction set, and a non-volatile recording medium such as an HDD or an SSD capable of recording an OS, an ○○ program, a DBMS (database server), etc. The communication unit 103 has an interface for physically connecting to a network, executes communication control with the network NW, and performs input / output of information.

[0071] Figure 2(b) is a hardware configuration diagram of the terminal device 90. As shown in Figure 2, the terminal device 90 includes a control unit 901, a storage unit 902, a communication unit 903, an input unit 904, and an output unit 905, which are used to exert the functions of each part and each process.

[0072] The control unit 901 has a processor such as a CPU that can execute an instruction set, and executes an OS, a program, etc. The storage unit 902 has a volatile memory such as a RAM that can store an instruction set, and a non-volatile recording medium such as an HDD or an SSD that can record an OS, a determination program, etc. The communication unit 903 has an interface for physically connecting to a network, executes communication control with the network NW, and performs input / output of information. The input unit 904 has an operation input device capable of input processing such as a touch panel or a keyboard, and an audio input device capable of audio input such as a microphone. The output unit 905 has a display device capable of display processing such as a display, and an audio output device such as a speaker.

[0073] <4. Functional configuration> FIG. 3 is a diagram showing the functional configurations of the support device 1, the user terminal 2, and the generation device 3 in the present embodiment. The support device 1 includes an acquisition unit 11, an evaluation unit 12, a specification unit 13, a generation unit 14, a proposal unit 15, and a database DB. This is specifically realized by information processing by software by hardware (such as the control unit 101). For example, the control unit 101 reads out a document correction program stored in the storage unit 102 and expands it in a temporary storage area (such as a RAM) of the storage unit 102 and executes it, and controls the communication unit 103 etc., whereby the functions of the support device 1 are realized.

[0074] Also, the user terminal 2 includes a display unit. The user terminal 2 receives the information transmitted by the proposal unit 15, and the display unit performs a display based on the information. A display example will be described later.

[0075] The acquisition unit 11 acquires item information corresponding to the missing items of the target document. The item information is information regarding the description content corresponding to the missing items. Specifically, it includes description examples indicating the specific description content corresponding to the items, explanations regarding the items, sentences or texts indicating the gist of the items, evaluation comments regarding the missing items of the target document, and the like. In the present embodiment, description examples registered in advance in association with the items are acquired from the database DB. The acquisition unit 11 also acquires auxiliary information. In the present embodiment, the user terminal 2 receives an input of auxiliary information from the user, and the acquisition unit 11 acquires the auxiliary information via the user terminal 2. In the present embodiment, the acquisition unit 11 acquires target document information indicating the content of the target document being created as the auxiliary information. Further, the acquisition unit 11 of the present embodiment acquires an item type indicating the type of the missing item and a correction intensity selected by the user.

[0076] The evaluation unit 12 acquires evaluation information indicating the evaluation result of the target document and passes it to the acquisition unit 11. The evaluation unit 12 of the present embodiment acquires the evaluation information by performing an evaluation based on the target document information acquired by the acquisition unit 11, but it may be configured to acquire the evaluation information from an external device or the user terminal 2. Details of the evaluation process will be described later.

[0077] The specification unit 13 specifies a description example to be used for creating or correcting the target document from among the description examples corresponding to the item type of the missing item in the target document from the database DB that stores a plurality of description examples in association with the item type. The specification unit 13 may further specify the description example based on the correction intensity selected by the user.

[0078] The generation unit 14 inputs the item information and auxiliary information acquired by the acquisition unit 11 into a sentence generation model and obtains the output thereof. The acquisition unit 11 of the present embodiment generates a generation command instructing to generate a text regarding the missing item based on the item information and auxiliary information as a sentence in natural language and transmits it to the generation device 3. The transmission of this generation command corresponds to the input of the item information and auxiliary information to the sentence generation model in the present invention. Details of the generation process of the generation command will be described later.

[0079] The proposal department 15 outputs proposal information for assisting in creating missing items in the target document based on the output of the text generation model for the generation command. In this embodiment, the output is performed by transmitting the proposal information to the user terminal 2. In this embodiment, the proposal information includes a text corresponding to the missing item generated by the text generation model, and is presented to the user by a proposal screen described later that is displayed on the user terminal 2.

[0080] The database DB stores a plurality of description examples in association with item types. In this embodiment, the description examples stored in the database DB are descriptions in legal documents created in the past during the negotiation process, and the descriptions of past legal documents are registered in association with item types. Note that the description examples in the present invention are not limited to legal documents actually created in the past negotiation process. For example, a template of the company's own legal document may be registered in the database DB as a description example corresponding to the item type.

[0081] The item type is a classification of items according to their gist, and is a type for specifying what each item defines. For example, as described in 1. Definition, in this embodiment, the arguments are taken as items. Therefore, in this embodiment, the type of argument or the argument is taken as the item type, and in the database DB, the description examples are registered in association with the type of argument or the argument, and each description example shows an example of the description content within the range related to each argument. When taking clauses as items, the content of each clause may be registered as a description example.

[0082] For example, for each contract type, it is possible to define the types of arguments that can be included in the contract. More specifically, for example, in the case of a "confidentiality agreement" contract type, groups such as exception provisions for confidential information, liability provisions for damages, anti-social clauses, etc. are assumed as the types of arguments. Here, the group corresponds to a group of clauses that define a coherent set of regulations. And each type of argument includes one or more individual arguments. For example, in the type of "exception provisions for confidential information", the matters specifically defined as exceptions are an example of an "argument". For each argument, multiple description examples may be registered.

[0083] Figure 4 is an example of the description example information stored in the database DB in this embodiment. The description example information includes a description example ID, a document ID, an item type, a modification intensity, and a description example. In this embodiment, the description example is the description content of the corresponding part of a legal document created in a past agreement for each item type, or a template according to the item type. The document ID is an ID indicating the document from which the description example is derived when the description example is a legal document created in a past agreement. When the description example is a template, an ID indicating that it is a template is stored as the document ID in the description example information. Note that the data structure can be arbitrarily changed.

[0084] In addition, information regarding the background of the document from which each description example is derived, such as the creation date and time of the document from which the description example is derived, the inputter who input the document into the database DB, the legal affairs person in charge of the document, the contract type, the contractor (the other party to the contract), the position of one's own company, the contract object, etc., which corresponds to the above background information, may be further stored. Also, such information may be stored separately linked to the document ID.

[0085] The contract type is an example of the type of agreement stipulated in the document. Other types of agreements include atypical contracts, regulations, or rules. The position of the company refers to the contractual position, such as the information of the discloser or recipient of information in a confidentiality agreement. The contract object is an example of the "object of the agreement" and refers to goods, services, or businesses that are the subject of each contract (agreement). For example, the subject goods in a sales contract correspond to the contract object.

[0086] Note that some of the above-mentioned components of the support device 1 and part of the database DB may be provided in a computer separate from the support device 1, such as the user terminal 2 or the generation device 3. Also, the support device 1 may be provided with a display unit or the generation device 3.

[0087] <5. Processing Procedure> Next, with reference to FIG. 5, the processing procedure for creating and modifying a target document using the support system 0 of the present embodiment will be described. FIG. 5 is a sequence diagram showing an example of the processing procedure by the support device 1, the user terminal 2, and the generation device 3 in the support system 0 of the present embodiment. Note that the processing shown here is an example, and omissions and changes may be made without departing from the gist of the present invention. In the following example, the case where both the creation of the first draft of the document and the creation of the amendment are performed according to the present invention will be described, but the present invention may perform only either the creation of the first draft or the amendment. Also, the creation of the first draft and the creation of the amendment do not necessarily need to be performed by the same model, and they may be created using different models respectively.

[0088] (1) Creation of the First Draft Steps S501 to S506 are processes related to the creation of the first draft of the target document, that is, the creation of the text based on auxiliary information not including the target document information. Note that this process is not essential, and the processes after step S507 may be performed using a draft of the target document created by the user himself / herself, a draft generated using an external machine learning model, a contract draft received from the other party with whom the contract is to be concluded, etc.

[0089] First, in step S501, the user terminal 2 receives input of auxiliary information from the user and transmits it to the support device 1. As the auxiliary information, for example, background information or the like is assumed. In step S501, since it is the stage of creating a first draft of the target document, the auxiliary information does not include target document information indicating the content of the target document. On the other hand, the auxiliary information transmitted in step S501 needs to include information for specifying the item type of the item (deficient item) to be created. For example, by having the user select the item type of the item (deficient item) to be created, the acquisition unit 11 may acquire information indicating the item type as part of the background information, or based on the background information, the support device 1 may specify the item type of the item to be created.

[0090] Specifically, for example, when a confidentiality agreement is concluded when presenting information to a business partner, the relationship with the other party is the "information discloser", the type of agreement is the "confidentiality agreement" which is a type of agreement, and the desire of the parties is the modification intensity such as "want a strong (advantageous in one's own position) agreement", etc., which are examples of the background information acquired in step S501. Also, it is preferable that the acquisition unit 11 acquires the name or title of the contracting party and the designations (A, B) in the contract document, etc. as definition information.

[0091] In the input of this auxiliary information, the form of input by the user selecting from options is typically assumed. In addition to this, the form in which the user freely describes in a sentence expressed in natural language and the auxiliary information is input in text may also be used. Also, these auxiliary information may be acquired by term sheet information indicating the content of the term sheet used in contract negotiation and question information based on the question content and its answer regarding the content to be described in the legal document.

[0092] Next, in step S502, first, the specifying unit 13 specifies, from among the description examples stored in the database DB, the description example corresponding to the item type of the lacking clause and to be used for creating the target document. When the modification intensity acquisition unit 11 acquires the modification intensity indicating the modification intensity as a request of the party, the specifying unit 13 may specify, as the description example to be used for creating the target document, the description example corresponding to the item type of the lacking clause and corresponding to the acquired modification intensity, from among the description examples stored in the database DB. The generation unit 14 generates a generation instruction for causing the text generation model to generate a text regarding the lacking clause, based on the specified description example and auxiliary information.

[0093] Since the text generation model of the present embodiment is a conversation type model that generates a response text by taking a natural language text as an input, the generation instruction is generated as a natural language text. More specifically, the generation unit 14 can generate a generation instruction by generating a text combining an instruction text prepared in advance, the text specified based on the auxiliary information, and the description example. Also, for example, an instruction text may be prepared in advance for each option of the auxiliary information and stored in the storage unit of the support device 1, the instruction text may be selected according to the contract type among the auxiliary information, and the generation unit 14 may generate, as a generation instruction, a text combining the instruction text, the text specified based on the auxiliary information, and the description example. When the modification intensity acquisition unit 11 acquires the modification intensity indicating the modification intensity as a request of the party, instead of or in addition to being used for specifying the description example, the generation unit 14 may further generate, as a generation instruction, a text combining words corresponding to the acquired modification intensity.

[0094] More specifically, when the auxiliary information is input by selection, first, the words corresponding to the selected auxiliary information are identified. For example, for each option of the auxiliary information, the corresponding words can be set in advance, so that the words corresponding to the selected auxiliary information can be identified. When including the words corresponding to the correction intensity in the generation instruction, the words corresponding to the correction intensity are also identified in the same way. Then, based on these words, the description examples, and the instruction text registered in advance in the storage unit of the support device 1, the generation unit 14 generates a generation instruction. On the other hand, when the auxiliary information is input as free-form text, a generation instruction may be generated using the input auxiliary information, the description examples, and the instruction text prepared in advance.

[0095] Here, as the words corresponding to the auxiliary information, it is preferable to store a character string describing the document creation rules according to the auxiliary information. For example, in a legal document, the legal meaning may be different when an exception is described in a single sentence of a regulation and when a proviso sentence is separately described after a single sentence. By combining words that specify such description method rules according to the auxiliary information, an appropriate legal document can be created. Among the multiple document creation rules corresponding to each auxiliary information, the support device 1 may determine the document creation rules to be included in the generation instruction based on the auxiliary information, or include multiple document creation rules in the generation instruction and include in the generation instruction the fact that an appropriate document creation rule based on the auxiliary information is used from the multiple document creation rules, so as to determine the document creation rules used by the text generation model.

[0096] As instructions, for example, sentences such as "Please create a document based on the following conditions and the following description examples." and "Input auxiliary information indicating the agreement of the contract and description examples. Based on the following rules, modify the description examples according to the auxiliary information and create an appropriate contract text." can be stored in advance in the storage unit of the support device 1. Alternatively, instructions can be prepared for each contract type as options for auxiliary information, and information corresponding to the auxiliary information can be included in the instructions, such as "Please create a confidentiality contract based on the following conditions and the following description examples." Then, by generating a sentence that continues with words and description examples based on the auxiliary information after the instructions, a generation instruction for instructing the creation of a draft can be generated for the sentence generation model.

[0097] When the generation instruction is generated, the process proceeds to step S503, and the generation unit 14 sends the generation instruction to the generation device 3. The transmission of the generation instruction to the generation device 3 in step S503 corresponds to the input to the sentence generation model of the description examples and the auxiliary information. When the generation device 3 receives the generation instruction, it inputs the generation instruction into the sentence generation model. As a result, the sentence generation model outputs a first draft of the contract as a text regarding the missing items (all items required for the target document) (step S504).

[0098] The generation device 3 sends the text output by the sentence generation model to the support device 1 in step S505. Then, in step S506, the proposal unit 15 sends the first draft of the target document as proposal information to the user terminal 2 based on the text output by the sentence generation model. Through the above processing, the user can obtain the first draft of the target document by inputting the auxiliary information.

[0099] (2) Acquisition of evaluation information Next, the process related to the acquisition of the evaluation information of the target document will be described. Steps S507 to S509 are processes related to the evaluation of the target document. This process is for acquiring evaluation information indicating the evaluation of the target document. In the present embodiment, as will be described below, the evaluation unit 12 evaluates the target document, but evaluation information may be acquired by having an external device evaluate the target document, or evaluation information may be acquired by receiving evaluation information together with auxiliary information from the user terminal 2. Note that steps S506 and S507 may be omitted, and the evaluation of the target document may be continuously executed without waiting for the user's instruction.

[0100] In step S507, the user terminal 2 transmits an evaluation request to the support device 1. The evaluation request includes auxiliary information. The auxiliary information transmitted here includes target document information including at least part of the content of the target document to be evaluated and background information. In the present embodiment, a file of a document describing a draft of a contract is uploaded as the target document information.

[0101] In the support device 1, the acquisition unit 11 receives the evaluation request, and the evaluation unit 12 executes the evaluation of the target document based on the target document information and the background information. The evaluation of the target document may be performed by any method such as evaluation by a machine learning model or rule-based evaluation. As an example, in the present embodiment, a method of evaluating clauses based on rules according to the background information is used.

[0102] In particular, in this embodiment, evaluation is performed from a plurality of predetermined perspectives using the contract type and the contractual position among the background information. Note that the contract type may be determined by user selection or may be estimated from the title of the text or the title of each clause in the target text. Further, in the evaluation, the title of each clause in the target document may be used. Specifically, evaluation is performed on the content of each clause and the relationship between the plurality of clauses. Here, the plurality of clauses refers to a group of a plurality of clauses that define a set of regulations. For example, within the group, the presence or content of Clause B is determined according to the content of Clause A, and further Clause C is confirmed based on the result. By stipulating that the evaluation process branches according to the content, the relationship between the plurality of clauses can be evaluated based on rules. The evaluation of the content of each clause can be performed, for example, based on rules according to whether a specific keyword is included or the relationship between specific keywords (such as whether keyword X is used affirmatively or negatively).

[0103] In the evaluation of the target document, evaluation is performed from perspectives such as whether there is any unnecessary description (or a clause, description, etc. that is disadvantageous to oneself from the relationship with the other party in the background information), whether there are additional perspectives (missing items) to be added, and whether the content of the clause is appropriate. For example, for missing items, based on an item template that shows a list of items prepared in advance for each contract type and contractual position, it is determined based on rules whether each item is included in the target document, and the items not included are determined as missing items. Note that it may be determined according to the rules for each item, or a plurality of items may be determined according to one rule. In this embodiment, in particular, an item template is prepared for each contract type, and evaluation is performed on unnecessary descriptions and missing items according to the item template corresponding to the contract type.

[0104] In this embodiment, by registering in advance in the storage unit of the support device 1 the matters to be pointed out in case of violation of the above rules in association with the rules, the matters to be pointed out for the target document are identified by checking the rules. Specifically, for each of the above-described evaluation rules, the matters to be pointed out, the types of points to be pointed out, the evaluation comments, and the risk levels are stored in advance in the storage unit. The types of points to be pointed out in this embodiment are three types: "lack of items" which means the lack of items (arguments) to be included in the target document, "revision of clauses" which means the insufficiency of the description content of the clauses, and "items to be deleted" which means the existence of unnecessary descriptions, clauses or arguments. Further, the type of "lack of items" is further divided into "partial lack" where some of the clauses in a group defining a set of regulations are lacking, and "total lack" where the entire group is lacking.

[0105] The evaluation comment is a sentence or passage in natural language that proposes revisions or points out problems regarding the content of the target document. For example, it can be a sentence or passage such as "Don't you need to stipulate the negotiation clauses?" or "Isn't the period for assuming the confidentiality obligation set too long?" In addition, further, an explanatory sentence regarding the clause related to the point to be pointed out and information on an example description corresponding to the argument related to the point to be pointed out may be included in the evaluation comment. Also, the risk level is information indicating the degree of risk assumed in each matter to be pointed out. The risk level may be fixed (unchangeable) on the system side, or may be made changeable by user input.

[0106] Examples of setting the risk level are as follows. For pointed-out matters related to issues that are likely to occur after the contract and are related to important rights and obligations of the company, the risk level can be set to "A". For example, pointed-out matters regarding "ownership of intellectual property", "scope of damages", "definition of confidential information", etc. are candidates for pointed-out matters with a risk level of "A". Also, for pointed-out matters related to issues that are unlikely to cause problems even if omitted, the risk level can be set to "C". For example, it includes "exceptions to confidential information", "obligation to report during the process", etc. And for pointed-out matters that do not fall under A and C among multiple groups, the risk level can be set to "B". In addition to the above examples, the risk level may be set in advance for each argument, or it may be a risk determined each time based on the language of the clause. For example, the evaluation unit 12 may determine whether the company is advantaged or disadvantaged based on the language of the clause, and the risk level may be set based on the determination.

[0107] As described above, the evaluation unit 12 executes the evaluation process to generate evaluation information indicating the evaluation of the target document. The evaluation information includes, for each of the one or more pointed-out matters identified in the above process, the clause in the target document corresponding to the pointed-out matter, the evaluation comment, and the risk level. Also, when obtaining the evaluation information from an external device or the user terminal 2, evaluation information including the clause in the target document corresponding to the pointed-out matter, the evaluation comment, and the risk level is obtained for each pointed-out matter.

[0108] Note that the target document 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 the text generation model used for correction. For example, the acquisition unit 11 acquires target document information and background information, and the evaluation unit 12 reads out an evaluation rule for target document evaluation stored in advance based on the background information, generates an evaluation instruction based on the target document information and the evaluation rule corresponding to the background information, and transmits it to the generation device 3, and acquires the evaluation result (evaluation information) of the target document generated by the generation device 3. The evaluation instruction is generated as a sentence in natural language. As a result, it is possible to expect an effect of reducing the influence on the evaluation due to fluctuations in the notation of the sentence, differences in expressions, etc. Also, among a plurality of evaluation rules, it is possible to expect to perform an appropriate evaluation by using an evaluation rule (for example, an evaluation rule according to an item template set for each contract type) read out based on background information (for example, contract type). Note that the evaluation rule may be a rule for determining from the content of the target sentence which of the predetermined pointed-out matters should be given to the target document, or an evaluation rule for generating a sentence as the pointed-out matter to be given to the target document from the standard for each predetermined pointed-out matter and the content of the target document.

[0109] Furthermore, both the evaluation by the text generation model and the rule-based evaluation may be used in combination. At this time, when the evaluation by the text generation model and the rule-based evaluation are different, it is preferable to prioritize the rule-based evaluation. Thereby, even if uncertainty in the evaluation by machine learning occurs, a minimum evaluation can be guaranteed based on the rules. Note that the support device 1 may select an evaluation rule to be included in the generation instruction based on the background information from among a plurality of evaluation rules corresponding to each background information. Also, it may be configured such that a plurality of evaluation rules are included in the evaluation instruction, and an evaluation instruction includes a statement indicating that an appropriate rule is used based on the background information from among the plurality of rules, so as to determine the evaluation rule used by the text generation model.

[0110] When the evaluation unit 12 finishes evaluating the target document, in step S509, the evaluation unit 12 transmits the evaluation information to the user terminal 2. When the user terminal 2 receives the evaluation information, the evaluation information is subjected to display processing and the processing result is displayed on the display.

[0111] FIG. 6 shows a display example of an evaluation screen displayed on the user terminal 2 as a display processing result of the evaluation information. The evaluation screen W6 displays an overview of the target document and an overview of the evaluation result at the upper part of the screen. The content of the target document specified by the target document information is displayed on the left side of the screen. Also, a button for "editing the read content" is arranged, and when there is an error in the content of the target document information, the content can be edited by selecting this button.

[0112] On the right side of the screen, the evaluation result based on the evaluation information is displayed. For each pointed-out item, a pointed-out type indicating which of the unnecessary item, insufficient item, and content correction the pointed-out is (the display of "omission" and "correction" in FIG. 6) and an evaluation comment including a description example and explanation of the clause corresponding to the pointed-out item are displayed.

[0113] Also, for each pointed-out item, a pointed-out mark W61 of a different shape is attached for each pointed-out type, and a sequential number for each type is displayed. The same pointed-out mark W61 is also displayed together with the sequential number at the corresponding part of the pointed-out item in the target document on the left side of the screen, so that the user can recognize the position corresponding to the pointed-out item in the target document, and the target part in the target document is highlighted and displayed. In the example of FIG. 6, a circle mark indicates the correction of the clause, and a triangle mark indicates the pointed-out item related to the insufficient item.

[0114] Also, the pointed-out mark W61 is displayed in different colors according to the risk level. By operating the risk filter operation unit W62, the pointed-out items to be displayed according to the risk level can be filtered. In FIG. 6, a state is displayed in which it is set to display pointed-out items with a risk level of "B" or higher, that is, A or B.

[0115] In addition, a batch correction button W63 and an individual correction button W64 are displayed together with the evaluation results. When the user selects these buttons, a correction request for step S510 in FIG. 5 is transmitted. Depending on the selected button, the scope of the correction instruction is different.

[0116] There are three types of batch correction buttons W63: "Create batch amendments for all items", "Create amendments for item A", and "Create amendments for items A and B". When "Create batch amendments for all items" is selected, a correction request for all pointed-out matters is transmitted. When "Create amendments for item A" is selected, a correction request for the pointed-out matters with a risk level of A is transmitted. When "Create amendments for items A and B" is selected, a correction request for the pointed-out matters with a risk level of A and the pointed-out matters with a risk level of B is transmitted. Also, when the individual correction button W64 displayed for each pointed-out matter is selected, a correction request for that pointed-out matter is transmitted.

[0117] In addition, an input requesting batch correction may be accepted according to the type of pointed-out matter. For example, for pointed-out matters of item shortage, an input instructing the creation of the text of the missing items collectively may be accepted, or an input instructing the correction of clauses collectively may be accepted.

[0118] Also, in this embodiment, when the batch correction button W63 or the individual correction button W64 is selected, further requests from the parties related to the correction are accepted. For example, for pointed-out matters where multiple correction policies are assumed, the user can specifically select one of them to accept the request related to the correction. Alternatively, when a correction request is made for multiple pointed-out matters using the batch correction button W63, for example, a configuration may be adopted that allows the user to select the correction intensity such as "aggressive (greatly advantageous to the company)", "neutral", "conservative (more advantageous to the company than before but more advantageous to other companies)" as a whole. At this time, an instruction to generate only the amendment (when the type of pointed-out matter is "clause correction") or the text (when the type of pointed-out matter is "item shortage") corresponding to the correction intensity selected by the user may be generated, or an instruction to generate amendments or texts for all correction intensities may be generated. Also, such requests may be input in free-form text.

[0119] Here, when using the batch correction button W63 to correct some items such as "Create Amendment for Item A" and "Create Amendment for Items A and B", or when using the individual correction button W64 to make a correction request for some of the pointed-out matters, the correction request only includes the matters related to the pointed-out matters (item information and target document information for the corresponding part). As a result, when correcting some items such as Item A or Items A and B, it is even shorter than when correcting all items (the entire target document) at once, and when correcting one item by specifying one of the pointed-out matters. The correction instruction becomes shorter. In addition, by correcting some parts individually, it is possible to make it less likely to cause consistency problems associated with correcting multiple pointed-out matters, for example, when a single sentence in the target document contains multiple pointed-out matters. Furthermore, by allowing the user to select and correct the parts that are individually necessary, the processing time can be shortened, and the time required for correction can be reduced. Due to these effects, it is expected to improve the user experience. Therefore, it is more preferable to obtain item information and auxiliary information for the parts corresponding to some of the multiple pointed-out matters, especially the pointed-out matters selected by the user, and perform corrections.

[0120] (3) Correction Processing Next, the correction processing of the target document will be described. When receiving a correction request as described above, the acquisition unit 11 obtains target document information and evaluation information from the user terminal 2 as auxiliary information. As the target document information, the entire content of the target document may be obtained, or only the content of the clause corresponding to the previous text or the pointed-out matter that is the target of the correction request and the clauses before and after it may be obtained. The clause corresponding to the pointed-out matter is, for example, the clause related to the missing item when the pointed-out type is item shortage, and the clause to be corrected when the pointed-out type is clause correction. Especially when the pointed-out type is item shortage, the target document information includes information indicating the insertion position of the missing item in the target document and the description content before and after the insertion position.

[0121] More specifically, when there is a partial shortage in the pointed-out type, the description content of other clauses belonging to the same group as the clause regarding the shortage item is acquired as the target document information. Also, when the pointed-out type is a clause modification, if the clause to be modified is included in the group of any clause, it is preferable that the description content of other clauses belonging to the same group is included in the target document information. Further, the information indicating the insertion position is information indicating the place where the shortage item should be inserted in the target document, for example, information such as "between Clause A and Clause B". At this time, the content of the description before and after the insertion position is, for example, the content of Clause A and the content of Clause B. Note that the determination of the insertion position is determined in advance for items related to the shortage item and insertion rules (for example, under Clause A related to Clause B when Clause B is missing, etc.), and the insertion position is determined according to the related items and insertion rules. Note that the determination of the insertion position may be based on an item template showing a list of items prepared for each contract type or contractual position to determine the shortage items and sufficient items (items determined to be included in the target document), and the insertion position may be determined by an insertion rule based on the positional relationship on the lists of the shortage items and sufficient items (for example, determining the insertion position based on the list position of the shortage clause and the list position of the sufficient clause satisfying the condition, etc.).

[0122] Moreover, the acquisition unit 11 of the present embodiment further acquires background information and the modification intensity transmitted together with the modification request in step S510. In the present embodiment, the background information received from the user terminal 2 in step S501 or step S507 and the evaluation information generated by the evaluation unit 12 in step S508 are stored in the storage unit, and the acquisition unit 11 further acquires the evaluation information and the background information based on them. When the creation and evaluation of the target document are performed by an external device, the acquisition unit 11 may acquire the target document information and the evaluation information from the user terminal 2 or the external device.

[0123] Also in step S511, the acquisition unit 11 acquires the description examples from the database DB. The description examples used for creating the text regarding the amendment of the clause or the missing items are selected from the description examples registered in the database DB based on the item type of the argument corresponding to the pointed-out matters. Specifically, the acquisition unit 11 first identifies the item type of the missing items based on the evaluation information. Then, the identification unit 13 identifies, from the database DB, the description example that matches the amendment strength acquired by the acquisition unit 11 among the description examples corresponding to the item type. Then, the acquisition unit 11 acquires that description example from the database DB. Alternatively, the evaluation comments included in the evaluation information may be acquired by the acquisition unit 11 as item information. The evaluation comments here are sentences or texts in natural language that propose amendments or point out problems regarding the content of the target document, and may not include description examples.

[0124] Then, the generation unit 14 generates a generation command for causing the text generation model to generate an amendment to the target document or a text for the missing items based on the description example (item information), the target document information, the evaluation information, the background information, and the information on the requirements of the parties regarding the amendment. Since the text generation model of the present embodiment is a conversation type model that generates a response text by taking a text in natural language as an input, the generation command is generated as a text in natural language.

[0125] The generation unit 14 of the present embodiment generates a command to be transmitted to the generation device 3 according to the content of the pointed-out matters related to the evaluation request in step S507. When the pointed-out type of the pointed-out matters is the amendment of the clause, a generation command is generated to instruct to generate an amendment related to the pointed-out matters based on the target document information indicating the content of the part corresponding to the pointed-out matters in the target document and the background information. Then, by transmitting this to the generation device 3, the amendment is acquired as an output from the generation device 3 with the target document information and the background information as inputs. In addition to this, description examples may be further used as inputs.

[0126] On the one hand, when the type of the pointed-out item is lack of items, a generation instruction is generated to instruct to generate a text regarding the lacking items based on item information, for example, a description example specified by the specific part 13, and auxiliary information. Then, by transmitting this to the generation device 3, an amendment is obtained as an output from the generation device 3 with the description example and the auxiliary information as inputs. Here, as the auxiliary information, part or all of the target document information, background information, style information, definition information, and evaluation information can be used.

[0127] Specifically, as the generation instruction, a sentence is generated that describes an instruction regarding the generation of a text regarding the amendment or lacking items for each pointed-out item. However, for example, when there are a plurality of pointed-out items in a group of a plurality of clauses that define one clause or a set of regulations, if generation instructions are generated in units of pointed-out items, problems such as a plurality of conflicting amendments or texts being generated may occur. Therefore, the generation unit 14 of the present embodiment generates a sentence of the generation instruction for each clause or group of clauses corresponding to the plurality of pointed-out items for the generation instruction regarding a clause or group of clauses in which there are a plurality of pointed-out items.

[0128] The generation of the generation instruction will be specifically described below. First, regarding the background information, when the background information is input by selection, first, the words corresponding to the selected background information are specified. For example, by presetting the corresponding words for each option of the background information in advance, the words corresponding to the selected background information can be specified.

[0129] Similarly, regarding the desired information, when it is input by selection, first, the words corresponding to the selected desire are specified. For example, by presetting the corresponding words for each combination of the target pointed-out item and the options of the desire in advance, the words corresponding to the selected desire can be specified for each pointed-out item.

[0130] Then, based on those words, the target document information, the description example (item information), and the instruction text registered in advance in the storage unit of the support device 1, the generation unit 14 generates a generation command. On the other hand, when the background information and the request are input as free-form text, a generation command may be generated using the description example (item information), the input background information, and the instruction text prepared in advance.

[0131] Here, similar to the generation of the generation command in the creation of the first draft of the document, it is preferable to store a character string describing the rules for document creation according to the background information as the word corresponding to the background information. For example, in a legal document, the legal meaning may be different depending on whether an exception is described within a single provision sentence or a proviso sentence is separately described after a single sentence. By combining words that specify the rules of such description methods according to the background information, an appropriate legal document can be created. Among the plurality of rules corresponding to each background information, the support device 1 may determine the rules to be included in the generation command based on the background information, or include a plurality of rules in the generation command and include in the generation command that an appropriate rule is to be used from the plurality of rules based on the background information, so that the rules used by the document generation model can be determined.

[0132] The instruction text preferably varies according to the type of pointing out. For example, as an instruction text regarding the amendment of a clause, sentences such as "Regarding the above draft, the following points have been pointed out as a result of the review. Based on the following conditions, please create an amendment considering the points." and "Regarding the above draft, the following points have been pointed out as a result of the review. Based on the points, please create an appropriate contract document sentence referring to the following description example." can be stored in advance in the storage unit of the support device 1. Then, a sentence combining the description example, the target document information, the instruction text, and the words based on the background information or evaluation information may be generated as a generation command for instructing the document generation model to generate an amendment to the clause.

[0133] For example, as an instruction text regarding insufficient clauses, texts such as "The above draft lacks the following points. Based on the following conditions, refer to the description examples and create a text for the lacking points." and "The following points were pointed out as a result of the review of the above draft. Based on the points, refer to the following description examples and create a text for the clauses with the added lacking points." can be stored in advance in the storage unit of the support device 1. Furthermore, it is preferable to prepare different instruction texts for partial insufficiency and total insufficiency respectively. Then, a sentence combining the description example (item information), the target document information, the instruction text, and the words based on the background information and evaluation information may be generated as a generation instruction for instructing the sentence generation model to generate a sentence regarding the lacking clause.

[0134] When the generation instruction is generated, the process proceeds to step S512, and the generation unit 14 transmits the generation instruction to the generation device 3. Note that the generation instructions for each pointed-out matter, each clause, or each group of clauses may be transmitted separately multiple times. For example, it may be configured to first transmit the part of the instruction text in the generation instruction and then transmit the words according to the pointed-out matters and requests. Also, it may be configured to determine whether to transmit once or in multiple parts according to the number of characters of the generation instruction.

[0135] When the generation device 3 receives the generation instruction, it inputs the generation instruction into the sentence generation model. As a result, the sentence generation model outputs a text for the amendment of the clauses of the target document or the lacking items (step S513). Note that the generation unit 14 may generate a generation instruction for instructing to propose texts for a plurality of amendments or lacking items for one pointed-out matter in step S511.

[0136] Furthermore, a generation instruction may be sent to generate a reason for amendment based on the evaluation information, and in addition to the amendment or the text, the reason for amendment may be obtained. Further, the reason for amendment may be generated and obtained such that there are a reason for amendment for the company itself (or a reason for amendment for its own client) and a reason for amendment for the other party. In contract negotiation, the company and the other party repeat amendments to each other, and at this time, it is often the case that at least either the reason for amendment for internal explanation or the reason for amendment for external explanation is described. Although a high level of expertise is required for the description of the reason for amendment, the above configuration can assist even a user without specialized knowledge to be able to describe the reason for amendment.

[0137] In step S514, the generation unit 14 acquires the amendment or the text generated by the generation device 3. Then, the proposal unit 15 generates proposal information based on the amendment or the text, and in step S515, transmits the proposal information to the user terminal 2. On the user terminal 2, a proposal screen based on the proposal information is displayed.

[0138] (4) Re-evaluation process Although omitted in FIG. 5, in the present embodiment, by repeating the evaluation information acquisition process in step S508 described above and the amendment process in steps S511 to S514, the quality of the proposal can be further improved.

[0139] The evaluation unit 12 of the present embodiment receives, from the generation device 3, an amendment or a text of the target document as an output of the text generation model for the generation instruction. Then, evaluation information indicating the evaluation of the entire target document when the amendment or the text is reflected is obtained, and it is determined whether or not the evaluation satisfies a preset criterion. In the present embodiment, the evaluation unit 12 performs re-evaluation and generates evaluation information as a result, but the content of the target document after amendment may be transmitted to an external evaluation device, and the evaluation information may be obtained from the evaluation device as the evaluation result.

[0140] When the evaluation by the evaluation unit 12 meets the criteria, it causes the proposal unit 15 to send proposal information. On the other hand, when the evaluation does not meet the criteria, the generation unit 14 generates a generation instruction instructing to modify the amendment or the text based on the information indicating the content of the amendment or the text, the evaluation information, the background information, and the described case. Then, by the generation unit 14 sending the generation instruction to the text generation model, an amendment to the amendment or the text is further generated by the text generation model.

[0141] The evaluation unit 12 of this embodiment is configured to repeat the acquisition and evaluation of the amendment until the evaluation of the amendment meets a predetermined criterion or meets an end condition. As the predetermined criterion, for example, it can be set that the risk level for each of the following pointed-out matters included in the evaluation information is below a predetermined risk level. Also, as the end condition, for example, setting an upper limit on the number of repetitions or a lower limit on the number of characters changed by the correction is assumed. Hereinafter, the processing related to the repetition of the acquisition and correction of the evaluation information will be described with reference to the flowchart of FIG. 7.

[0142] First, when the support device 1 receives an amendment or a text that is the output of the text generation model, the evaluation unit 12 acquires this amendment or text in step S701.

[0143] Then, re-evaluation is executed in step S702. Specifically, re-evaluation is performed on the amendment in the same procedure as the above-described evaluation process. And the evaluation information indicating the evaluation of the amendment is generated in the same manner as in step S508. Here, in the re-evaluation, not the entire target document but only the part related to the previous pointed-out matters may be targeted. Specifically, the clauses or groups of clauses indicated by the amendment can be evaluated as the target of the re-evaluation, and the re-evaluation can be omitted for other parts. Thereby, an effect of improving the processing speed is expected.

[0144] Next, it is determined whether or not the evaluation of the amendment in step S703 meets the pre-set criteria. In the present embodiment, the risk level of the pointed-out matters in the evaluation information is confirmed, and it is determined that the criteria are met when the risk levels of all the pointed-out matters are below a predetermined risk level.

[0145] If the criteria are not met, the process proceeds to step S705 to check whether or not the number of loop iterations (number of corrections) has reached the upper limit. If the upper limit has not been reached (in step S705, "Y"), the process proceeds to step S706, where the evaluation unit 12 generates and transmits a generation command to the generation unit 14 again, adds the loop iteration counter, and returns to step S701.

[0146] If the criteria are met in step S703, or if the number of corrections has reached the upper limit in step S705, the process proceeds to step S704, where the evaluation unit 12 causes the proposal unit 15 to generate proposal information. Then, the proposal unit 15 transmits the proposal information to the user terminal 2 (corresponding to step S515 in FIG. 5) and ends the process. Note that the above processing procedure is an example, and it may be arbitrarily changed, for example, by swapping the determination order of steps S703 and S705.

[0147] In principle, loop processing is performed until the pre-set criteria are met. However, if the number of repetitions becomes extremely large or if it becomes impossible to exit the loop, an increase in the processing load is a concern. Therefore, in the present embodiment, an upper limit number of loop iterations is defined as an end condition. Even if the re-evaluation does not meet the criteria, if the end condition is met, the repetition of the re-evaluation is ended, the loop processing is exited, and the process proceeds to step S515. In addition, as another end condition, for example, it may be set that the number of changed characters becomes a certain value or less when comparing the documents before and after the correction. This can prevent the situation where only minor corrections occur from being repeated.

[0148] (5) Proposal screen Next, with reference to FIG. 8, the proposed screen that is displayed on the user terminal 2 as the display processing result of the proposed information will be described. FIG. 8 is an example of the display of the proposed screen in the present embodiment. The proposed screen W8 of the present embodiment is displayed in a configuration similar to the evaluation screen W6 shown above, and the description of the pointing mark W81 and the risk filter operation unit W82 will be omitted because they are the same as the pointing mark W61 and the risk filter operation unit W62 in the evaluation screen W6.

[0149] The main difference from the evaluation screen W6 is mainly the display content on the right side of the screen, particularly the part related to "revision proposal". In the proposed screen W8, as a display based on the text regarding the amendment or the lacking clause included in the proposed information, a revision proposal W83 is included for each pointed-out matter. The revision proposal W83 displays the above-mentioned amendment and also displays it in a way that the text can be copied. Also, when the support program of the present invention is provided as a program to be add-on (add-in, plug-in) in the document editing software, a correction button may be arranged instead of the copy button. In this case, when the user operates the correction button, the correction by the amendment is directly performed on the document editing software.

[0150] Also, it is preferable to highlight the changed points from the draft with highlights or the like so that the user can easily recognize them. Thereby, the user can obtain the content and the text regarding the amendment or the lacking item for each pointed-out matter, and even a user who is not familiar with legal knowledge can perform the amendment of the legal document by himself / herself.

[0151] In addition, when a plurality of amendments or texts are obtained for the pointed-out matters, since a plurality of amendments or texts regarding the pointed-out matters are included in these proposed information, it is preferable to display a plurality of revision proposals W83 for the pointed-out matters on the proposed screen W8 so that the user can select them.

[0152] Note that the output method of the proposed information is not limited to the display of the above-mentioned proposed screen W8. For example, the proposal unit 15 may send a document format file for the entire target document to which amendments have been applied to at least some of the pointed-out matters to the user terminal 2. Also, the user may be allowed to select whether to reflect the amendment proposals displayed on the proposed screen W8, or which amendment proposals to reflect, and based on the result, the proposal unit 15 may further send a document format file in which the entire draft has been corrected to the user terminal 2.

[0153] As described above, according to the support system of the present embodiment, even for a user without specialized knowledge, it is possible to support the user in making corrections to the draft of a legal document such as a contract. Also, for the missing items, an appropriate text can be created based on the described plan. Furthermore, by re-evaluating the amendment or the text and issuing an order to create a new amendment according to the result, the quality of the target document correction can be improved.

[0154] Also, by generating an instruction in natural language text, it is possible to create an amendment or a text using a conversational text generation model that generates a response text based on the input text. As a result, it is possible to cause the model to generate an amendment or a text that takes into account various conditions such as the background and requirements of the contract, and provide an amendment or a text of the target document that better suits the user's situation and wishes.

[0155] Furthermore, a generation instruction may be sent to the text generation model so that the reason for the correction is also generated based on the evaluation information, and the reason for the correction may be included in the proposed information. Furthermore, the reason for the correction may be generated as a reason for correction for the company (or for the user's own client) and a reason for correction for the other party. In contract negotiations, the company and the other party repeatedly make corrections to each other. At this time, it is often necessary to describe at least one of the reason for the correction for internal explanation and the reason for the correction for external explanation. Although a high level of expertise is required to describe the reason for the correction, the above configuration can support even a user without specialized knowledge in describing the reason for the correction.

Explanation of Signs

[0156] 0: Support system 1: Support device 11: Acquisition unit 12: Evaluation unit 13: Identification unit 14: Generation unit 15: Proposal unit 2: User terminal 3: Generation device 10: Information processing device 101: Control unit 102: Memory unit 103: Communication unit 90: Terminal device 901: Control unit 902: Memory unit 903: Communication unit 904: Input unit 905: Output unit NW: Network W6: Evaluation screen W61: Indication mark W62: Risk filter operation unit W63: Batch correction button W64: Individual correction button W8: Proposal screen W81: Indication mark W82: Risk filter operation unit W83: Correction proposal

Claims

1. A support program for supporting the creation or modification of a target document, The subject document is a legal document to be created or amended; A computer is caused to function as an acquisition unit, a generation unit, and a suggestion unit; The acquisition unit is Item information on the contents of the document that correspond to the missing items, which are missing in the document; Acquire auxiliary information related to the target document for making a suggestion according to the target document; The generation unit inputs the item information and auxiliary information into a sentence generation model; The suggestion unit outputs suggestion information for assisting in the creation of the missing items in the target document based on the output of the sentence generation model.

2. The support program according to claim 1 , wherein the item information includes examples of legal documents corresponding to the missing items, and the examples are descriptions or templates in documents that prescribe past agreements.

3. The computer further functions as an evaluation unit; The evaluation unit acquires evaluation information indicating an evaluation result of the target document, The assistance program according to claim 1 , wherein the acquisition unit acquires the item information based on the evaluation information.

4. The evaluation information includes an item type indicating a type of the missing item, The support program according to claim 3 , wherein the acquisition unit acquires the example corresponding to the item type of the missing item from a database that stores examples showing the contents of the items in association with the item types.

5. The evaluation information includes evaluation comments regarding the deficiencies, The assistance program according to claim 3 , wherein the acquisition unit acquires the evaluation comments as the item information.

6. The evaluation information includes one or more points of criticism regarding the target document, The problem is assigned a classification selected from a plurality of groups including a missing item and a correction of a clause; The acquisition unit determines the indication type of the indication included in the evaluation information, For items that are found to be insufficient, obtain the item information of the insufficient items related to the items that are found to be insufficient; Regarding the matters to be amended, obtain target document information indicating the contents of the target document related to the matters to be amended; The generation unit is For items where there is a shortage of items, the item information and auxiliary information are input to the sentence generation model, 4. The support program according to claim 3, wherein, for a suggested item for amending a clause, the target document information and information on an agreement stipulated by the target document are input to the text generation model.

7. The target document includes a plurality of clauses; The types of indications indicating deficiencies are divided into several types for a group of multiple clauses that define a set of regulations, including partial deficiency, where some of the clauses in the group are missing, and total deficiency, where the entire group is missing. The support program according to claim 6 , wherein the generation unit determines an input to the sentence generation model depending on whether an indicated matter relating to a group of a plurality of clauses is partially insufficient or completely insufficient.

8. When the type of indication related to the item deficiency is the partial deficiency, The support program according to claim 7 , wherein the acquisition unit acquires, as the auxiliary information, contents of other clauses in the target document that belong to the same group as the clause relating to the missing item.

9. The support program according to claim 1 , wherein the auxiliary information includes target document information indicating at least a part of the content of the target document.

10. 10. The support program according to claim 9, wherein the target document information includes information on a location in the target document where the missing item is to be inserted, and information on contents written before and after the insertion location.

11. The target document includes a plurality of clauses; The support program according to claim 9 , wherein the target document information includes a description of a clause in the target document that is related to a missing item.

12. the subject document is a legal document including a preamble; The support program according to claim 9 , wherein the target document information includes a description of a preamble of the target document.

13. The support program according to claim 1 , wherein the auxiliary information includes definition information indicating definitions of words used in the target document.

14. The support program according to claim 1 , wherein the auxiliary information includes writing style information indicating a writing style in the target document.

15. The acquisition unit further acquires a correction strength selected by a user, The support program according to claim 1 , wherein the generation unit further inputs the correction strength into the sentence generation model.

16. causing the computer to further function as a specific unit; The acquisition unit further acquires an item type indicating a type of the missing item and a correction strength selected by a user, The identification unit identifies an example of writing to be used in creating or correcting the target document from among the examples of writing corresponding to the item type in a database that stores a plurality of examples of writing corresponding to the item type, based on the correction level; The support program according to claim 2 , wherein the generation unit inputs the example descriptions and auxiliary information identified by the identification unit into the text generation model.

17. The assistance program according to claim 1 , wherein the generation unit generates a generation command instructing to generate a draft sentence for the missing item based on the item information and auxiliary information as a sentence in natural language and inputs the generated command to the sentence generation model.

18. The computer further functions as an evaluation unit; The evaluation unit acquires evaluation information indicating an evaluation result of the target document, The evaluation information includes one or more points of criticism regarding the target document, The problem is assigned a classification selected from a plurality of groups including a missing item and a correction of a clause; The acquisition unit determines the indication type of the indication included in the evaluation information, For items that are found to be insufficient, obtain item information on the insufficient items related to the items that are found to be insufficient; Regarding the matters to be amended, obtain target document information indicating the contents of the target document related to the matters to be amended; The generation unit is Regarding the item lacking, the generation command instructing the sentence generation model to generate a draft sentence regarding the lacking item based on the item information and auxiliary information is input, The support program according to claim 17 , wherein, with respect to a suggested item for amending a clause, the generation command instructing the sentence generation model to generate a proposed amendment related to the suggested item based on the target document information and auxiliary information is input to the sentence generation model.

19. A support system for supporting the creation or modification of a target document, comprising: The subject document is a legal document to be created or amended; The apparatus includes an acquisition unit, a generation unit, and a proposal unit, The acquisition unit is Item information on the contents of the document corresponding to the missing items that are missing in the legal document to be created or amended; Acquire auxiliary information related to the target document for making a suggestion according to the target document; The generation unit inputs the example sentences and auxiliary information into a sentence generation model, The suggestion unit transmits suggestion information for assisting in the creation of the missing items in the target document based on an output of the sentence generation model.

20. A method for assisting in the creation or modification of a target document, comprising: The subject document is a legal document to be created or amended; The method is executed by a computer including an acquisition unit, a generation unit, and a suggestion unit, The acquisition unit is Item information on the contents of the document corresponding to the missing items that are missing in the legal document to be created or amended; Acquire auxiliary information related to the target document for making a suggestion according to the target document; The generation unit inputs the example sentences and auxiliary information into a sentence generation model, The suggestion unit transmits suggestion information for supporting the creation of the missing items in the target document based on an output of the sentence generation model.

Images