Information processing method

Abstract

To realize mitigation of workload on a user in a system that assists a user in reviewing a document.SOLUTION: An information processing method in an embodiment comprises: acquiring a correction policy for each of at least one input unit of a first legal document; acquiring at least one corrected input unit that has been corrected using at least one prompt that is generated on the basis of the at least an input unit and the correction policy for the input unit, and a natural language processing model; and outputting the at least one corrected input unit.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] FIELD OF THE INVENTION An embodiment of the present invention relates to an information processing method. [Background technology]

[0002] Conventionally, there are systems that support users in reviewing documents. The users revise the documents while referring to the document review results output by the system. This places a heavy burden on the users. There is room for further improvement in reducing the user's workload. Summary of the Invention [Problem to be solved by the invention]

[0003] An embodiment of the present invention has been made in view of the above circumstances, and aims to reduce the workload of a user in a system that supports the user in reviewing documents. [Means for solving the problem]

[0004] An information processing method according to an embodiment includes obtaining a correction policy for each of at least one input unit of a first legal document, obtaining at least one corrected input unit using at least one prompt generated based on at least the input unit and the correction policy for the input unit and a natural language processing model, and outputting the at least one corrected input unit. [Brief explanation of the drawings]

[0005] [Figure 1] FIG. 2 is a diagram showing an overview of an example of processing in the correction suggestion system 100 according to the embodiment. [Figure 2] FIG. 1 is a diagram showing the configuration of a correction suggestion system 100 according to an embodiment. [Figure 3] FIG. 10 is a diagram showing an example of data related to a revision policy. [Figure 4A] FIG. 1 is a block diagram showing an example of a hardware configuration of an information processing device 10 according to an embodiment. [Figure 4B] FIG. 2 is a block diagram showing an example of the hardware configuration of a terminal device T1 according to the embodiment. [Figure 5] FIG. 2 is a diagram showing an example of functional blocks in a processor 21. [Figure 6] FIG. 6 is a diagram showing an example of a document review screen 60 displayed as a web browser screen on the display device 30 of the terminal device T1 during document review processing. DETAILED DESCRIPTION OF THE INVENTION

[0006] An embodiment will be described below with reference to the drawings.

[0007] In this specification, the term "unit" does not simply mean a physical means, but also includes cases where the functions of the "unit" are realized by software. Furthermore, the functions of one "unit" or device may be realized by two or more physical means, devices, or software, and the functions of two or more "units" or devices may be realized by one physical means, device, or software.

[0008] FIG. 1 is a diagram showing an overview of an example of processing in a correction suggestion system 100 according to an embodiment.

[0009] The modification suggestion system 100 of this embodiment is an information processing system realized by a modification suggestion program, and is an information processing system that proposes modifications to the contents of target items in a contract, which is a document to be supported.

[0010] For example, when a contract is concluded, the parties to the contract reach an agreement. When the parties to the contract reach an agreement, one or both of the parties may inspect and amend the contract. For example, in the case of a contract between companies, the legal department in charge of contract inspection will inspect the contents of the contract and amend it as necessary. Note that contract inspection is also sometimes called confirmation, examination, or review.

[0011] The legal documents to be processed by the revision proposal system 100 of this embodiment are, for example, documents that contain legal content that generates predetermined legal effects within the document, and that include criteria based on predetermined viewpoints such as whether the legal content or formality is favorable or unfavorable. Examples of such documents include contracts, applications, memoranda, company regulations, policies, etc. In this embodiment, a contract will be used as an example. Furthermore, legal documents may also be referred to as documents to be supported, as appropriate.

[0012] The legal document for which revisions are proposed by the revision suggestion system 100 of this embodiment is computer-processable, such as an electronic document. This may be an electronic document created using predetermined electronic document creation software, or an electronic document obtained by digitizing a paper contract using a predetermined image analysis technology (e.g., OCR (Optical Character Recognition) technology).

[0013] The revision suggestion system 100 of this embodiment provides a document creation support service. The document creation support service is a service that includes at least one of the following: executing a document review process on a support target document on a web browser and providing the result to the user; automatically correcting the document using the result of the document review process and presenting the result as a revision candidate (revision suggestion process); editing the document in response to an instruction from the user; saving the edited document or providing it to the user; editing additional information associated with the support target document (e.g., meta information, comment information, chat information, format information, etc.); and creating a new document (including a copy).

[0014] In particular, the revision suggestion process by the revision suggestion system 100 of this embodiment is executed using an artificial intelligence model that uses natural language. Here, the artificial intelligence model that uses natural language is a natural language processing model represented by, for example, GPT (Generative Pre-trained Transformer). This natural language processing model receives natural language text called a prompt as input and outputs natural language text that is automatically generated in response to the prompt.

[0015] As the natural language processing model used in the revision suggestion system 100 according to this embodiment, for example, the natural language processing models disclosed in https: / / arxiv.org / abs / 2203.02155 and https: / / openai.com / blog / chatgpt / can be used.

[0016] Furthermore, the revision suggestion system 100 of this embodiment automatically generates a prompt (a natural language string for querying the natural language processing model) to be input to the natural language processing model in the revision suggestion process, based on at least one input unit generated from the document to be supported and a revision policy for that input unit. Here, the input unit is, for example, a sentence in terms of an article, paragraph, or subsection, and can be generated by dividing the document to be supported according to a predetermined standard. The revision policy is information indicating the policy for revising the input unit, and is composed of natural language expressions such as, for example, "Delete the wording XX," "Must include the wording XX," "Must make this a mandatory provision," and "Must make this an advisory clause."

[0017] The revision policy can be acquired, for example, by referring to rules that define multiple combinations of revision policies with text types (e.g., text types such as texts specifying purpose, texts specifying title, and texts specifying time periods). These rules are created by, for example, an expert such as a lawyer and managed in a database. The combinations of text types and revision policies managed in the database can be added or modified at any time as needed. The revision policy can also be acquired, for example, by a trained model that uses an input unit as input and a revision policy as output. It can also be acquired by combining processing using a trained model with rule-based processing using a database. For example, the database contains information in which multiple sample sentences are associated with multiple revision policies. The revision suggestion system 100 may acquire a revision policy for each of at least one input unit based on the similarity between at least one input unit and multiple sample sentences. The sample sentences may include a sentence before revision and a sentence after revision.

[0018] FIG. 1 illustrates an example of a correction suggestion process that includes an input unit generation process, a classification process, a prompt creation process as preprocessing, natural language processing, and a corrected sentence presentation process. In the input unit generation process, a sentence segmentation process is performed using, for example, a sentence classifier. In the classification process, the types of sentences obtained by the segmentation process are classified using, for example, a classifier. In the prompt creation process, a prompt including an input unit and a correction policy is created by, for example, referring to a database. In natural language processing, a corrected sentence is obtained using, for example, a prompt including an input unit and a correction policy and a natural language processing model. In the corrected sentence presentation process, the corrected sentence obtained by, for example, natural language processing is presented to the user as a correction candidate. That is, in natural language processing, multiple correction candidates are generated as candidates for at least one corrected input unit using at least one prompt and a natural language processing model. In the corrected sentence presentation process, the generated multiple correction candidates are output (presented). It is also possible to save multiple correction candidates and output the saved multiple correction candidates in response to a user instruction.

[0019] The natural language processing model used in natural language processing may be stored in an external information processing device 200 that is connected to the correction suggestion system 100 via a network and has at least one processor and memory, as shown in Figure 1. The natural language processing model may also be provided by a model vendor in the form of an API or the like. The external information processing device 200 may have, for example, the same configuration as the information processing device shown in Figure 4A.

[0020] In this specification, the document review process refers to, for example, a process of analyzing whether the information contained in the document to be supported is appropriate based on certain criteria and outputting the results. For example, the document review process includes checking and examining whether the information contained in the document to be supported is appropriate based on certain criteria. Specifically, the document review process includes comparing the wording in the document with a pre-prepared checklist and outputting the comparison results or the indications contained in the checklist.

[0021] In the correction suggestion process, for example, the input unit to be included in the prompt of the natural language processing model may be generated by dividing the entire text of the document to be supported into units such as articles, etc. Also, for example, the input unit to be included in the prompt of the natural language processing model may be generated for sentences in units of articles, paragraphs, or subparagraphs that have been determined to be inappropriate or should be corrected by the document review process.

[0022] In the correction suggestion process, the correction suggestion system 100 according to this embodiment obtains at least one corrected input unit using at least one prompt generated based on at least an input unit and a correction policy for the input unit and a natural language processing model.

[0023] Therefore, according to this embodiment, it is possible to impose constraints based on the correction policy on the natural language processing model, and to obtain corrected input units that conform to the correction policy. As a result, it is possible to obtain corrected sentences that are more specific and accurate than those obtained in natural language processing that uses as input a prompt that does not include a correction policy. In contrast, in natural language processing that uses as input a prompt that does not include a correction policy, the corrected sentences obtained often end up being highly dependent on the training data of the natural language processing model used.

[0024] A user receiving the document creation support service of this embodiment can review and consider, on a browser screen, the revised document as a candidate for the revision proposal automatically generated in the revision proposal process. Furthermore, the user can revise the document to be supported based on the review results simply by confirming the displayed revision proposal or selecting the desired revision proposal. In particular, even when revising legal documents such as contract documents, which require specialized knowledge, it is possible to obtain candidates for revisions that are highly likely to be in line with the revision guidelines and revise the document to be supported by selecting them. As a result, the user's workload in revising the document to be supported can be significantly reduced.

[0025] 2 is a diagram showing the configuration of an embodiment of a correction suggestion system 100. The correction suggestion system 100 is, for example, an information processing system that is communicably connected to a terminal device T1 via a network such as the Internet. Details of the correction suggestion system 100 will be described later.

[0026] The terminal device T1 is, for example, a computer used by a user of the correction suggestion system 100, such as a smartphone, a tablet terminal, or a personal computer.

[0027] For example, a user accesses the amendment proposal system 100 through the terminal device T1 and uploads a contract, which is an example of a legal document, to the amendment proposal system 100. The user then receives amendment proposal information for a target item in the contract through the terminal device T1. In this specification, the term "contract" can be appropriately interpreted as a "legal document."

[0028] Although only one terminal device T1 is shown in FIG. 2, the number of terminal devices T1 is not limited to this.

[0029] Next, details of the revision proposal system 100 will be described. The revision proposal system 100 of this embodiment includes a storage unit 110, a contract acquisition unit 120, an input unit generation unit 130, a revision policy information generation unit 150, a revision proposal input information (prompt) generation unit 160, and a revision proposal unit 170. Each unit shown in Fig. 2 can be realized, for example, by using a storage area or by a processor executing a program stored in the storage area.

[0030] The storage unit 110 of this embodiment stores information processed in the revision suggestion system 100. The storage unit 110 can store, for example, contract information, input unit information, and policy information, which will be described later. The storage unit 110 may be, for example, a memory.

[0031] The contract acquisition unit 120 of this embodiment, for example, acquires contract information related to a contract and stores it in the storage unit 110. The contract acquisition unit 120 can acquire electronic data of a contract.

[0032] Here, the contract information is information relating to the contract, and includes, for example, status information, position information, contract type information (collectively referred to as meta information), and contract content information.

[0033] The status information in this embodiment is information indicating the status of the contract, for example, information indicating the revision status of the contract. Examples of status information include "Partner's Draft" indicating that the contract is a draft (for example, a first draft) from the other party to the contract, "Partner's Revision Proposal" indicating that the contract has been revised by the other party to the contract, or "Reviewed" indicating that the contract has been confirmed by the company (the business entity to which the user belongs).

[0034] The position information in this embodiment is information indicating the position of a contract, for example, information indicating the position of your company (the business to which the user belongs). The position information may be, for example, "outsourcing party" indicating the outsourcing party in a business outsourcing contract, or "disclosing party" indicating the party disclosing confidential information in a confidentiality agreement.

[0035] The contract type information in this embodiment is information indicating the type of contract, and may be, for example, a "service outsourcing contract" or a "non-disclosure agreement."

[0036] The contract content information in this embodiment is information (for example, character data) that indicates the content of the contract.

[0037] The contract acquisition unit 120 of this embodiment may acquire contract information from, for example, the terminal device T1. That is, the user may provide (upload) a contract to the modification proposal system 100 through the terminal device T1, and the contract acquisition unit 120 may acquire the contract information of the uploaded contract.

[0038] At this time, the contract acquisition unit 120 may acquire contract information from the user, for example, through the operation of the terminal device T1. In this case, when receiving an upload of a contract from the terminal device T1, the contract acquisition unit 120 may acquire the contract information by accepting input (for example, input from a keyboard or selection of an item preset by the system) of at least a portion of the contract information (for example, status information, position information, contract type information) from the user.

[0039] Furthermore, the contract acquisition unit 120 may acquire contract information from, for example, another information processing system. In this case, for example, the user may provide contract information to the other information processing system via the terminal device T1, and the contract acquisition unit 120 may acquire the contract information provided from the other information processing system. Examples of other information processing systems include a contract management system, a document management system, and cloud storage.

[0040] Furthermore, the contract acquisition unit 120 may analyze the content information of the contract and acquire at least a portion of the contract information (for example, status information, position information, contract type information).

[0041] The contract information stored in the storage unit 110 includes, for example, a contract ID, a sequence ID, status information, position information, contract type information, and contract content information.

[0042] The contract ID in this embodiment is information that identifies a contract processed in the amendment suggestion system 100. The contract ID is identification information that is assigned so that the contract before and after amendment of the same contract can be identified, that is, the contract ID is information that uniquely identifies the contract (i.e., electronic document) acquired by the contract acquisition unit 120.

[0043] The sequence ID in this embodiment is information that identifies the contract indicated by the contract processed in the revision proposal system 100. In other words, the sequence ID is a unit for managing, for example, a series of contracts that have been drafted, revised, and agreed upon (i.e., multiple contracts, such as a draft contract, a contract under revision, and an agreed upon contract, aimed at concluding the contract).

[0044] The contract acquisition unit 120 of this embodiment may generate a sequence ID, for example, in response to a user's operation of the terminal device T1. Specifically, when a user uploads a contract before amendment, for example, the contract before amendment is uploaded to an area for managing contracts (for example, a virtual folder managed on the amendment suggestion system 100), and a specific sequence ID (for example, a sequence ID corresponding to the virtual folder or a sequence ID identical to the sequence ID of the first contract before amendment) may be generated.

[0045] The sequence ID of this embodiment can be used, for example, as follows. During negotiations over the contents of a contract, the other party may propose amendments to the user's draft contract, and the user may wish to further amend the proposed amendments. The contract or draft contract is sent by the other party, for example, by email. However, deleting the other party's amendments without modification may be undesirable from a negotiation perspective, even though it may improve the contract's content. In such cases, by referencing the contents of a contract having the same sequence ID (i.e., the previous version), the user can input amendment guidelines that are more compromising than those applied to the previous version, soften the contract's wording, or generate more polite comments to be presented along with the proposed amendments. Furthermore, by having information about the previous version of the contract under review, the user can reduce the risk of rehashing previous compromises, which could lead to difficult negotiations.

[0046] In addition, if there are contracts related to the contract under review, such as a contract previously concluded before the contract under review was renewed, a basic contract if the contract under review is an individual contract, or a joint research and development contract and a joint patent application contract, which stipulate the same or related content in two or more contracts, or if there are contracts with the same counterparty, the content of those contracts and the status of negotiations may be taken into consideration.

[0047] These may be taken into consideration at the stage of generating the modification policy, or at the stage of generating a prompt, which will be described later, but it is preferable to generate them as a prompt, as this simplifies the process.

[0048] The contract acquisition unit 120 of this embodiment may extract information based on, for example, the contracting parties in the contract, the date the contract was created, and the contract information of the contract (for example, status information, position information, contract type information, contract content information).

[0049] The input unit generation unit 130 of this embodiment divides the contents of the contract into items included in the contract based on the contract information, and stores the input units in the storage unit 110.

[0050] Here, the input unit may be, for example, the entire contract, or a unit obtained by dividing the contract into constituent units such as the title, heading, body, signature, etc., or may correspond to the clauses contained in the contract. In other words, the input unit generation unit 130 can divide the contents of the contract, for example, by "article" or "paragraph."

[0051] For example, the input unit generation unit 130 can analyze the contents of a contract and divide the contents of the contract into items based on the positions of characters corresponding to the input units (e.g., the characters for "article" or "paragraph"). The input unit generation unit 130 does not need to divide all contracts into input units, nor does it need to divide all items of a contract into items. That is, the input unit generation process by the input unit generation unit 130 may only be performed on a portion of a contract or only on a part of a contract. The input unit generation unit may also combine multiple divided units or some of the multiple divided units into one as needed. For example, one legal issue may be defined across multiple articles. In this case, if the contract is divided into articles as described above and a revision policy or revision proposal is obtained or proposed for each article, it may be difficult to obtain an appropriate revision policy or propose a revision. In such cases, it is preferable to combine parts having similar legal issues so that an appropriate revision policy can be obtained using rules or a trained model.

[0052] Furthermore, prior to these processes, information required for document review, such as text data, may be extracted from the received document data, or processing such as OCR may be performed to extract text data from image data.

[0053] When the contract acquisition unit 120 acquires the contract information of a contract, the input unit generation unit 130 can divide the contract before amendment into items and generate input unit information indicating the content before amendment for each input unit.

[0054] The input unit information stored in the storage unit 110 includes, for example, an input unit ID, a contract ID, and input unit information.

[0055] The input unit ID in this embodiment is information that identifies item information processed in the revision suggestion system 100. The input unit information in this embodiment is information that indicates the content corresponding to the input unit, for example, information that indicates the content of each clause.

[0056] The revision policy information generating unit 150 of this embodiment generates information regarding a policy on how to preferably revise a contract. For example, a document review process is performed on the contract to generate document review information. For example, the document review information is based on information obtained by analyzing the document based on rules or standards applied to the document (e.g., certain regulations, laws, or customs). For example, contract review information such as that of this embodiment may include: (1) indication of whether clauses contained in the document (in this embodiment, the contract) are advantageous or disadvantageous to the user (the parties); (2) advice on modifying or deleting clauses contained in the contract; (3) indication of missing items that should normally be included in the contract and suggestions for adding items; and (4) information on the importance or recommended level of each review result item. Formatting information, such as whether the text format meets standards and whether specified terminology is used, may also be included.

[0057] The revision policy information generating unit 150 of this embodiment executes a document review process using contract data (an example of data of a first legal document) received from the terminal device T1. That is, the analyzing unit 151 generates document review information (described later) for the document data as part of the document review process. The document review information is an example of analysis information.

[0058] The analysis unit 151 of this embodiment is included in the revision policy information generation unit 150 and extracts, for example, information necessary for generating revision policy information that is included in a document (document data). Various techniques can be adopted as a method for extracting information from document data. For example, a keyword search of the document data may be performed to extract alphanumeric characters, etc., that are included near the keyword. Information may also be extracted by morphological analysis, syntactic analysis, semantic analysis, or contextual analysis. Furthermore, information may be extracted by natural language processing using machine learning. Note that revision policy information generation processing may be performed on a received document without extracting information necessary for generating policy information as described above.

[0059] The revision policy information generating unit 150 can use various techniques to generate document review information. For example, it is preferable to use the extracted information to generate document review information based on a rule base that is based on rules designed by humans, particularly lawyers who are experts. Alternatively, a trained model can be used that receives the extracted information as input and outputs corresponding document review information. Furthermore, a combination of a rule-based approach and a trained model can be used.

[0060] The revision policy information generating unit 150 can also generate document review information according to the position of the user (the party) by using meta information associated with the contract. Furthermore, the analyzing unit 103 can also generate document review information by referring to the contents of other documents (other versions of the support target document, other contracts related to the contract, etc.). Meta information includes information added to the document file itself, such as information about the document creator, and information added by the user, etc., by being input separately from the file.

[0061] The meta information in this embodiment includes, for example, at least one of the following: document ID, document type (e.g., basic transaction agreement, non-disclosure agreement, outsourcing contract, etc.), version, creation date and time, update date and time, upload date and time, creator, updater, deadline, and document ID of related document.

[0062] The revision policy information generating unit 150 of this embodiment generates a prompt. The prompt is based on, for example, a sentence in an input unit and a revision policy instruction based on document review information for the input unit. Specifically, the revision policy information generating unit 150 generates a prompt written in natural language, such as "This input is (the clause divided into input units)" or "Instruction is (revision policy)."

[0063] The prompt thus obtained is input to a natural language processing model, and a corrected input unit is obtained by correcting the input unit.

[0064] The natural language processing model may be stored in an information processing device connected to the system via a network, or may be provided by the model vendor in the form of an API or the like.

[0065] 4A is a block diagram showing an example of a hardware configuration of an information processing device 10 according to an embodiment. As shown in FIG. 4A, the information processing device 10 is a computer, and includes, for example, a processor 11, a main storage device (memory) 13, an auxiliary storage device (memory) 15, a network interface 17, a device interface 18, and a bus 19 connecting the respective components.

[0066] Although the information processing device 10 shown in FIG. 4A includes one of each component, it may include multiple of the same component. Also, while FIG. 4A shows one information processing device 10, the software may be installed on multiple computers, and each of the multiple computers may execute the same or different parts of the software. In this case, a form of distributed computing may be used in which each computer communicates with the other computers via a network interface 17 or the like to execute the processing. In other words, the information processing device 10 in this embodiment may be configured as a system in which one or multiple computers execute instructions stored in one or multiple storage devices to realize various functions described below.

[0067] Various calculations of the information processing device 10 may be executed in parallel using one or more processors, or using multiple computers via a network. Furthermore, various calculations may be distributed to multiple processing cores within a processor and executed in parallel. Furthermore, some or all of the processes, means, etc. disclosed herein may be executed by at least one of a processor and a storage device provided on a cloud that can communicate with the information processing device 10 via a network. Thus, the processing of the document creation support system S in this embodiment may be in the form of parallel computing using one or more computers.

[0068] The processor 11 may be, for example, an electronic circuit (processing circuit, processing circuitry, CPU (Central Processing Unit), GPU (Graphics Processing Unit), FPGA (Field Programmable Gate Array), ASIC (Application Specific Integrated Circuit), etc.) including a control device and an arithmetic device of the information processing device 10. The processor 11 may also be a semiconductor device including a dedicated processing circuit. The processor 11 is not limited to an electronic circuit using electronic logic elements, but may also be realized by an optical circuit using optical logic elements. The processor 11 may also include an arithmetic function based on quantum computing.

[0069] The processor 11 performs arithmetic processing based on data and software (programs) input from each device, etc., configured internally of the information processing device 10, and can output arithmetic results and control signals to each device, etc. The processor 11 may control each component constituting the information processing device 10 by executing an OS (Operating System), applications, etc. of the information processing device 10.

[0070] The information processing device 10 in this embodiment may be realized by one or more processors 11. Here, the processor 11 may refer to one or more electronic circuits arranged on one chip, or may refer to one or more electronic circuits arranged on two or more chips or two or more devices. When multiple electronic circuits are used, the electronic circuits may communicate with each other via wire or wirelessly.

[0071] The main memory device 13 is a memory device that stores, for example, instructions executed by the processor 11 and various data, and information stored in the main memory device 13 is read by the processor 11. The auxiliary memory device 15 is a memory device other than the main memory device 13. Note that these memory devices refer to any electronic component capable of storing electronic information, and may be semiconductor memory. The semiconductor memory may be either volatile memory or non-volatile memory. The memory device for saving various data used in the information processing device 10 according to this embodiment may be realized by the main memory device 13 or the auxiliary memory device 15, or may be realized by an internal memory built into the processor 11. For example, the memory unit in this embodiment may be realized by the main memory device 13 or the auxiliary memory device 15.

[0072] A plurality of processors may be connected (coupled) to one storage device (memory), or a single processor 11 may be connected. A plurality of storage devices (memories) may be connected (coupled) to one processor. When the information processing device 10 in this embodiment is configured with at least one storage device (memory) and a plurality of processors connected (coupled) to this at least one storage device (memory), it may include a configuration in which at least one of the plurality of processors is connected (coupled) to at least one storage device (memory). This configuration may also be realized by storage devices (memories) and processors 11 included in a plurality of computers. Furthermore, it may include a configuration in which the storage device (memory) is integrated with the processor 11 (for example, a cache memory including an L1 cache and an L2 cache).

[0073] The network interface 17 is an interface for connecting to the network N, for example, wirelessly or by wire. The network interface 17 may be an appropriate interface, such as one that conforms to an existing communication standard. Information may be exchanged with an external device connected via the network N through the network interface 17. The network N may be any one of a WAN (Wide Area Network), a LAN (Local Area Network), a PAN (Personal Area Network), etc., or a combination thereof, as long as information is exchanged between the information processing device 10 and the external device. An example of a WAN is the Internet, an example of a LAN is IEEE802.11 or Ethernet (registered trademark), and an example of a PAN is Bluetooth (registered trademark) or NFC (Near Field Communication), etc.

[0074] The device interface 18 is an interface such as a USB (Universal Serial Bus) that directly connects to an output device such as a display device, an input device, or an external device. The external device may also be a storage device such as a storage device (memory), network storage, or HDD. The external device may also be a device that has some of the functions of the components of the information processing device 10 in this embodiment. In other words, the information processing device 10 may transmit or receive some or all of the processing results of the external device.

[0075] Fig. 4B is a block diagram showing an example of the hardware configuration of terminal device T1 according to the embodiment. As shown in Fig. 4B, terminal device T1 is a computer, and includes, for example, processor 21, main storage device (memory) 23, auxiliary storage device (memory) 25, network interface 27, device interface 28, bus 29 connecting the respective components, display device 30 connected to device interface 28, and input device 31 connected to device interface 28. Note that the specific hardware configurations of processor 21, main storage device 23, auxiliary storage device 25, network interface 27, device interface 28, and bus 29 are similar to those of processor 11, main storage device 13, auxiliary storage device 15, network interface 17, device interface 18, and bus 19, respectively, and therefore description thereof will be omitted.

[0076] The display device 30 is a display device such as an LCD (Liquid Crystal Display), a CRT (Cathode Ray Tube), a PDP (Plasma Display Panel), an organic EL (Electro Luminescence) panel, etc. Based on instructions from the processor 21, the display device 30 displays, for example, a document review screen and a document editing screen, which will be described later.

[0077] The input device 31 is, for example, a device such as a keyboard, a mouse, or a touch panel. A user can use the input device 31 to input various editing instructions, for example, in a document editing process described below. The instructions input from the input device 31 are sent to the processor 21.

[0078] The terminal device T1 may also include other external devices (for example, a camera, a microphone, a motion capture device, various sensors, etc.) connected via the device interface 18 in order to manipulate and edit document data.

[0079] 5 is a diagram showing an example of functional blocks in the processor 21. The processor 21 includes, for example, a second control unit 40 and a browser unit 42 as functions implemented by the processor 21. The functions implemented by the second control unit 40 and the browser unit 42 are stored as programs in, for example, the main storage device 23 or the auxiliary storage device 25. The processor 21 can implement the functions related to the second control unit 40 and the browser unit 42 by reading and executing the programs stored in the main storage device 23 or the auxiliary storage device 25.

[0080] The second control unit 40 starts up the browser unit 42 in response to, for example, an instruction from the user.

[0081] The browser unit 42 is, for example, a general web browser, and executes processing in accordance with HTML (Hyper Text Markup Language) data, scripts (for example, JavaScript (registered trademark)), etc. When the browser unit 42 is started, it transmits an initial access URL to the information processing device 10 and accesses it.

[0082] The browser unit 42 of this embodiment displays data including text data and image data received from the information processing device 10 in a predetermined format on the display device 30 of the terminal device T1. More specifically, in a document review process, the browser unit 42 displays the document review information, document data, additional information, etc. received from the information processing device 10 in a predetermined format based on layout information on the display device 30 of the terminal device T1. Furthermore, in a document editing process, the browser unit 42 of this embodiment displays the document review information, editing data, additional information, etc. received from the information processing device 10 in a predetermined format based on layout information on the display device 30 of the terminal device T1.

[0083] 6 is a diagram showing an example of a document review screen 60 displayed as a web browser screen on the display device 30 of the terminal device T1 during the document review process. As shown in Fig. 6, on the document review screen 60 of this embodiment, the support target document is displayed in a document display area 61, additional information such as a revision history and comments is displayed in an additional information area 67, and document review information is displayed in a document review area 62.

[0084] For example, an edit button 68 for instructing the start of document editing processing is displayed in the document display area 61. Next to the additional information area 67 where additional information is displayed, a scroll bar 64 for scrolling up and down the support target document displayed in the document display area 61 and the additional information displayed in the additional information area 67 is displayed.

[0085] The user clicks on the button displayed on the screen to start the document review process, and issues an instruction to start the document review process. The system accesses the database, performs the document review process, and obtains the revision policy for each input unit. Figure 3 shows an example of a database related to revision policies.

[0086] At this time, all revision policies set for the type of input unit (for example, the type of clause) may be obtained, or only a portion of the relevant revision policies for the input unit of the document to be supported may be obtained, but from the perspective of convenience for review, it is preferable to display only a portion of the revision policies.

[0087] Next, a prompt for acquiring a corrected input unit is generated based on the input unit of the document to be supported and the corresponding correction policy. The prompt is in natural language, such as "Revise Article A in accordance with the content of Point A." The prompt is input to a natural language processing model, and the corrected input unit generated based on the prompt is acquired as output. If the natural language processing model is located outside the system, the system transmits and receives data to and from an information processing device having the natural language processing model using a network such as the Internet.

[0088] The natural language processing model may generate a prompt after the review process, before the user instructs the acquisition of a corrected input unit, such as by clicking button 69 that instructs the generation of a corrected input unit, or after the user instructs the generation of a corrected input unit. The "correction policy" in boxes such as 623 in FIG. 6 corresponds to the correction policy in this specification, and the "corrected sentence example" in box 623 corresponds to the "sample sentence" in this specification. The prompt may also be generated based on optional correction policies. The optional policies may include making the text of the corrected output unit as close as possible to the text of the input unit, making the corrected output unit more favorable to the user than the correction policy, or omitting part of the correction policy from the corrected output unit. These optional policies may be displayed. The user can select a desired policy from the displayed optional policies.

[0089] A button 69 may be provided for each of the multiple input units or correction policies to indicate the generation of a corrected input unit. This allows the user to obtain corrected input unit information for the input unit that the user considers necessary. This is advantageous when, for example, the generation of corrected input units is charged per unit, and the number of units to be generated needs to be adjusted.

[0090] The system then sends information including the above-described corrected input unit to the user. The system applies the proposed amendment sentence to the input unit or the contract automatically, or in accordance with a user instruction, such as clicking a button 70 indicating that the proposed amendment sentence should be applied to the draft contract. For example, the system replaces an input unit in the document to be supported displayed in the browser with the corrected input unit. Before or after the replacement, the user can edit the corrected input unit or instruct the system to generate another corrected input unit. The acquired corrected input unit and desired instructions may also be input into a natural language processing model to obtain a more preferable corrected input unit. Furthermore, the system may generate at least one corrected input unit for each natural language processing model using at least one prompt and multiple different natural language processing models, and output the at least one corrected input unit for each natural language processing model. Furthermore, the system may store the at least one corrected input unit generated for each natural language processing model, and output the stored at least one corrected input unit in response to a user instruction.

[0091] A button 70 may be provided for indicating whether or not to apply the modified input information to each of the multiple modified input units. This allows the user to check the modified input information and apply the modified input unit information to the input units that the user deems necessary. Furthermore, before outputting at least one modified input unit, it may be checked whether or not the acquired at least one modified output unit satisfies modified output criteria including the modification policy.

[0092] The support target document of this embodiment may be associated with the following comment information and chat information.

[0093] In this embodiment, the comment information associated with a document is additional information created by at least one user involved in the creation of the document, and is associated and managed for each version of the document, for example. The comment information associated with a document is typically information separate from comments created by word processing software, but can also include comments created by word processing software as necessary.

[0094] In this embodiment, chat information refers to additional information generated by multiple users involved in creating a document using a chat function, and is managed in association with multiple versions of the document. This additional information is typically text data, but can also include other data formats such as images and audio.

[0095] For example, the user can save the edited contract. The saved contract can be used by printing it out as a Word or PDF file, downloading it, sending it from the system, or sending a URL.

[0096] This embodiment has at least the following configuration. (1) obtaining a correction policy for each of the at least one input unit of the first legal document; obtaining at least one corrected input unit using at least one prompt generated based on at least the input unit and the correction policy for the input unit and a natural language processing model; outputting said at least one modified input unit; An information processing method comprising: (2) The correction policy is acquired by referring to a database. The information processing method described in (1). (3) The acquisition of the correction policy is determining a class for each of the at least one input unit with a classifier; the database includes information in which a plurality of types are associated with a plurality of modification policies; using the determined type and the database; The information processing method according to (2) above, (4) referencing the database to obtain the correction policy and sample sentences for each of the at least one input unit; generating the prompt based on the sample sentences obtained; The information processing method according to (2) above, (5) the database includes information in which a plurality of sample sentences are associated with a plurality of revision strategies; obtaining the correction policy for each of the at least one input unit based on a similarity between the at least one input unit and a plurality of the sample sentences; The information processing method according to (4) above, (6) The sample sentences include sentences before and after correction. (5) The information processing method described in (5). (7) generating a plurality of candidate corrections for the at least one corrected input unit using the at least one prompt and the natural language processing model; outputting a plurality of said correction candidates; The information processing method according to (1), comprising: (8) storing a plurality of said correction suggestions; outputting the saved plurality of correction candidates in response to a user instruction; The information processing method according to (7) above, (9) using the at least one prompt and a plurality of different natural language processing models to generate the at least one modified input unit for each of the natural language processing models; outputting the modified at least one input unit for each of the natural language processing models; The information processing method according to (1), comprising: (10) storing the modified at least one input unit generated for each of the natural language processing models; outputting the saved modified at least one input unit in response to a user instruction; The information processing method according to (9) above, (11) the first legal document is a contract; The information processing method described in (1). (12) The contract was sent by the other party to the contract. (11) The information processing method according to (11). (13) The at least one input unit is any one of an article, a paragraph, and a subsection. The information processing method described in (1). (14) before outputting the at least one modified input unit, checking whether the acquired at least one modified input unit satisfies modified output criteria including the modification policy; The information processing method described in (1). (15) transmitting information displaying a button that allows the input unit of the first legal document to be replaced with one corrected input unit; The information processing method described in (1). (16) The prompts are further generated based on the option policy of the modification. The information processing method described in (1). (17) The option policy includes any one of making the text of the corrected input unit as close as possible to the text of the input unit, making the corrected input unit have content that is more advantageous to the user than the correction policy, and making the corrected input unit omit part of the correction policy. (16) The information processing method according to (16). (18) The option policy is displayed so that the user can select it. (16) The information processing method according to (16).

[0097] Although the embodiments of the present disclosure have been described in detail above, these embodiments are presented as examples and are not intended to be limited to individual embodiments. Each embodiment can be modified by various additions, changes, substitutions, partial deletions, combinations, etc. These embodiments and their modifications are included in the scope of the invention described in the claims and their equivalents.

Claims

[Claim 1] obtaining a correction policy for each of the at least one input unit of the first legal document; obtaining at least one corrected input unit using at least one prompt generated based on at least the input unit and the correction policy for the input unit and a natural language processing model; outputting said at least one modified input unit; An information processing method comprising:

Images